Hit enter to search or ESC to close

Paper Digest: ICLR 2021 Highlights

admin

Download ICLR-2021-Paper-Digests.pdf– highlights of all ICLR-2021 papers. Readers can also choose to read this highlight article on our console, which allows users to filter out papers using keywords.

The International Conference on Learning Representations (ICLR) is one of the top machine learning conferences in the world. In 2021, it is to be held online. In the following highlight list, paper 1-53 correspond to oral presentations; paper 54-167 correspond to spotlight presentations; paper 168-860 correspond to poster presentations. Different from all other conferences, some ICLR papers come with author-created one-sentence summaries. We keep such summaries as highlights if they are available and provide machine generated highlights otherwise.

If you do not want to miss any interesting academic paper, you are welcome to sign up our free daily paper digest service to get updates on new papers published in your area every day. You are also welcome to follow us on Twitter and Linkedin to get updated with new conference digests.

Paper Digest Team
team@paperdigest.org

TABLE 1: Paper Digest: ICLR 2021 Highlights

Paper Author(s)
1 What Matters for On-Policy Deep Actor-Critic Methods? A Large-Scale Study
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We conduct a large-scale empirical study that provides insights and practical recommendations for the training of on-policy deep actor-critic RL agents.		 Marcin Andrychowicz; Anton Raichuk; Piotr Stanczyk; Manu Orsini; Sertan Girgin; Rapha�l Marinier; Leonard Hussenot; Matthieu Geist; Olivier Pietquin; Marcin Michalski; Sylvain Gelly; Olivier Bachem;
2 Theoretical Analysis of Self-Training with Deep Networks on Unlabeled Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper provides accuracy guarantees for self-training with deep networks on polynomial unlabeled samples for semi-supervised learning, unsupervised domain adaptation, and unsupervised learning.		 Colin Wei; Kendrick Shen; Yining Chen; Tengyu Ma;
3 Learning to Reach Goals Via Iterated Supervised Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose and formally analyze a simple RL method for learning goal-reaching policies via iterated supervised learning.		 Dibya Ghosh; Abhishek Gupta; Ashwin Reddy; Coline Manon Devin; Benjamin Eysenbach; Sergey Levine;
4 Deep Symbolic Regression: Recovering Mathematical Expressions from Data Via Risk-seeking Policy Gradients
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A deep learning approach to symbolic regression, in which an autoregressive RNN emits a distribution over expressions that is optimized using a risk-seeking policy gradient.		 Brenden K Petersen; Mikel Landajuela Larma; Terrell N. Mundhenk; Claudio Prata Santiago; Soo Kyung Kim; Joanne Taery Kim;
5 Optimal Rates for Averaged Stochastic Gradient Descent Under Neural Tangent Kernel Regime
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This is the first paper to overcome technical challenges of achieving the optimal convergence rate under the NTK regime.		 Atsushi Nitanda; Taiji Suzuki;
6 Free Lunch for Few-shot Learning: Distribution Calibration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We proposed a simple but effective few-shot classification method, which can estimate an unseen class distribution from one sample. Code is available at: https://github.com/ShuoYang-1998/ICLR2021-Oral_Distribution_Calibration		 Shuo Yang; Lu Liu; Min Xu;
7 Scalable Learning and MAP Inference for Nonsymmetric Determinantal Point Processes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose scalable learning and maximum a posteriori (MAP) inference algorithms for nonsymmetric determinantal point processes (DPPs).		 Mike Gartrell; Insu Han; Elvis Dohmatob; Jennifer Gillenwater; Victor-Emmanuel Brunel;
8 Randomized Automatic Differentiation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a general framework and approach for randomized automatic differentiation (RAD), which can allow unbiased gradient estimates to be computed with reduced memory in return for variance.		 Deniz Oktay; Nick McGreivy; Joshua Aduol; Alex Beatson; Ryan P Adams;
9 Learning Generalizable Visual Representations Via Interactive Gameplay
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show the representation learned through interaction and gameplay generalizes better compared to passive and static representation learning methods.		 Luca Weihs; Aniruddha Kembhavi; Kiana Ehsani; Sarah M Pratt; Winson Han; Alvaro Herrasti; Eric Kolve; Dustin Schwenk; Roozbeh Mottaghi; Ali Farhadi;
10 Global Convergence of Three-layer Neural Networks in The Mean Field Regime
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a rigorous framework for three-layer neural networks in the mean field regime and prove a global convergence guarantee.		 Huy Tuan Pham; Phan-Minh Nguyen;
11 Rao-Blackwellizing The Straight-Through Gumbel-Softmax Gradient Estimator
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We reduce the variance of the straight-through Gumbel-Softmax estimator to improve its performance.		 Max B Paulus; Chris J. Maddison; Andreas Krause;
12 Rethinking Attention with Performers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce Performers, linear full-rank-attention Transformers via provable random feature approximation methods, without relying on sparsity or low-rankness.		 Krzysztof Marcin Choromanski; Valerii Likhosherstov; David Dohan; Xingyou Song; Andreea Gane; Tamas Sarlos; Peter Hawkins; Jared Quincy Davis; Afroz Mohiuddin; Lukasz Kaiser; David Benjamin Belanger; Lucy J Colwell; Adrian Weller;
13 Getting A CLUE: A Method for Explaining Uncertainty Estimates
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a method to help explain uncertainties of any differentiable probabilistic model by perturbing input features.		 Javier Antoran; Umang Bhatt; Tameem Adel; Adrian Weller; Jos� Miguel Hern�ndez-Lobato;
14 When Do Curricula Work?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We conduct extensive experiments over thousands of orderings to investigate the effectiveness of three kinds of learning: curriculum, anti-curriculum, and random-curriculum.		 Xiaoxia Wu; Ethan Dyer; Behnam Neyshabur;
15 Federated Learning Based on Dynamic Regularization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present, FedDyn, a novel dynamic regularization method for Federated Learning where the risk objective for each device is dynamically updated to ensure the device optima is asymptotically consistent with stationary points of the global loss.		 Durmus Alp Emre Acar; Yue Zhao; Ramon Matas; Matthew Mattina; Paul Whatmough; Venkatesh Saligrama;
16 Geometry-aware Instance-reweighted Adversarial Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper has proposed a novel adversarial training method, i.e., geometry-aware instance-reweighted adversarial training (GAIRAT), which sheds new lights on improving the adversarial training.		 Jingfeng Zhang; Jianing Zhu; Gang Niu; Bo Han; Masashi Sugiyama; Mohan Kankanhalli;
17 Co-Mixup: Saliency Guided Joint Mixup with Supermodular Diversity
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new perspective on joint mixup augmentation and formulate the optimal construction of a batch of mixup data.		 JangHyun Kim; Wonho Choo; Hosan Jeong; Hyun Oh Song;
18 SenSeI: Sensitive Set Invariance for Enforcing Individual Fairness
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new invariance-enforcing regularizer for training individually fair ML systems.		 Mikhail Yurochkin; Yuekai Sun;
19 End-to-end Adversarial Text-to-Speech
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Efficient, adversarially trained feed-forward text-to-speech model producing high-quality speech, learnt end-to-end in a single stage.		 Jeff Donahue; Sander Dieleman; Mikolaj Binkowski; Erich Elsen; Karen Simonyan;
20 Dataset Condensation with Gradient Matching
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper proposes a training set synthesis technique that learns to produce a small set of informative samples for training deep neural networks from scratch in a small fraction of computational cost while achieving as close results as possible.		 Bo Zhao; Konda Reddy Mopuri; Hakan Bilen;
21 Rethinking Architecture Selection in Differentiable NAS
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an alternative perturbation-based architecture selection that directly measures each operation’s influence on the supernet.		 Ruochen Wang; Minhao Cheng; Xiangning Chen; Xiaocheng Tang; Cho-Jui Hsieh;
22 A Distributional Approach to Controlled Text Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel approach to Controlled Text Generation, relying on Constraints over Distributions, Information Geometry, and Sampling from Energy-Based Models.		 Muhammad Khalifa; Hady Elsahar; Marc Dymetman;
23 Learning Cross-Domain Correspondence for Control with Dynamics Cycle-Consistency
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We learn correspondence across domains with different modalities, physics parameters, and morphologies for control tasks (in both simulation and real robot) with Dynamics Cycle-Consistency.		 Qiang Zhang; Tete Xiao; Alexei A Efros; Lerrel Pinto; Xiaolong Wang;
24 Human-Level Performance in No-Press Diplomacy Via Equilibrium Search
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an agent that approximates a one-step equilibrium in no-press Diplomacy using no-regret learning and show that it exceeds human-level performance		 Jonathan Gray; Adam Lerer; Anton Bakhtin; Noam Brown;
25 Parrot: Data-Driven Behavioral Priors for Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a method for pre-training a prior for reinforcement learning using data from a diverse range of tasks, and use this prior to speed up learning of new tasks.		 Avi Singh; Huihan Liu; Gaoyue Zhou; Albert Yu; Nicholas Rhinehart; Sergey Levine;
26 Invariant Representations for Reinforcement Learning Without Reconstruction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our goal is to learn representations that provide for effective downstream control and invariance to task-irrelevant details.		 Amy Zhang; Rowan Thomas McAllister; Roberto Calandra; Yarin Gal; Sergey Levine;
27 Do 2D GANs Know 3D Shape? Unsupervised 3D Shape Reconstruction from 2D Image GANs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Unsupervised 3D Shape Reconstruction from 2D Image GANs		 Xingang Pan; Bo Dai; Ziwei Liu; Chen Change Loy; Ping Luo;
28 Varying Coefficient Neural Network with Functional Targeted Regularization for Estimating Continuous Treatment Effects
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a varying coefficient network and a functional targeted regularization for estimating continuous treatment.		 Lizhen Nie; Mao Ye; qiang liu; Dan Nicolae;
29 Rethinking The Role of Gradient-based Attribution Methods for Model Interpretability
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Input-gradients in discriminative neural net models capture information regarding an implicit density model, rather than that of the underlying discriminative model which it is intended to explain.		 Suraj Srinivas; Francois Fleuret;
30 Neural Synthesis of Binaural Speech
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an end-to-end approach to neural binaural sound synthesis that for the first time outperforms DSP-based methods in a qualitative evaluation and in a perceptual study. We will release a first-of-its-kind binaural audio dataset as a benchmark for future research.		 Alexander Richard; Dejan Markovic; Israel D. Gebru; Steven Krenn; Gladstone Alexander Butler; Fernando Torre; Yaser Sheikh;
31 DiffWave: A Versatile Diffusion Model for Audio Synthesis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: DiffWave is a versatile diffusion probabilistic model for waveform generation, which matches the state-of-the-art neural vocoder in terms of quality and can generate abundant realistic voices in time-domain without any conditional information.		 Zhifeng Kong; Wei Ping; Jiaji Huang; Kexin Zhao; Bryan Catanzaro;
32 An Image Is Worth 16×16 Words: Transformers for Image Recognition at Scale
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Transformers applied directly to image patches and pre-trained on large datasets work really well on image classification.		 Alexey Dosovitskiy; Lucas Beyer; Alexander Kolesnikov; Dirk Weissenborn; Xiaohua Zhai; Thomas Unterthiner; Mostafa Dehghani; Matthias Minderer; Georg Heigold; Sylvain Gelly; Jakob Uszkoreit; Neil Houlsby;
33 On The Mapping Between Hopfield Networks and Restricted Boltzmann Machines
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Hopfield networks with correlated patterns can be mapped to Restricted Boltzmann Machines with orthogonal weights.		 Matthew Smart; Anton Zilman;
34 SMiRL: Surprise Minimizing Reinforcement Learning in Unstable Environments
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Using Bayesian surprise as an unsupervised intrinsic reward function to learn complex behaviors in unstable environments.		 Glen Berseth; Daniel Geng; Coline Manon Devin; Nicholas Rhinehart; Chelsea Finn; Dinesh Jayaraman; Sergey Levine;
35 Evolving Reinforcement Learning Algorithms
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We meta-learn RL algorithms by evolving computational graphs which compute the loss function for a value-based model-free RL agent to optimize.		 John D Co-Reyes; Yingjie Miao; Daiyi Peng; Quoc V Le; Sergey Levine; Honglak Lee; Aleksandra Faust;
36 Growing Efficient Deep Networks By Structured Continuous Sparsification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an efficient training method that dynamically grows and prunes neural network architectures.		 Xin Yuan; Pedro Henrique Pamplona Savarese; Michael Maire;
37 Deformable DETR: Deformable Transformers for End-to-End Object Detection
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Deformable DETR is an efficient and fast-converging end-to-end object detector. It mitigates the high complexity and slow convergence issues of DETR via a novel sampling-based efficient attention mechanism.		 Xizhou Zhu; Weijie Su; Lewei Lu; Bin Li; Xiaogang Wang; Jifeng Dai;
38 EigenGame: PCA As A Nash Equilibrium
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We formulate the solution to PCA as the Nash of a suitable game with accompanying algorithm that we demonstrate on a 200TB dataset.		 Ian Gemp; Brian McWilliams; Claire Vernade; Thore Graepel;
39 Augmenting Physical Models with Deep Networks for Complex Dynamics Forecasting
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new principled framework for combining physical models with deep data-driven networks, for which we provide theoretical decomposition guarantees.		 Vincent LE GUEN; Yuan Yin; J�r�mie DONA; Ibrahim Ayed; Emmanuel de Bezenac; Nicolas THOME; patrick gallinari;
40 Complex Query Answering with Neural Link Predictors
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show how to answer complex queries by answering their sub-queries via neural link predictors, aggregating results via t-norms and t-conorms, and identifying the optimal variable substitutions by solving an optimisation problem.		 Erik Arakelyan; Daniel Daza; Pasquale Minervini; Michael Cochez;
41 Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Coding
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our work addresses key issues in disentanglement research for moving towards more natural settings.		 David A. Klindt; Lukas Schott; Yash Sharma; Ivan Ustyuzhaninov; Wieland Brendel; Matthias Bethge; Dylan Paiton;
42 Self-training For Few-shot Transfer Across Extreme Task Differences
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Self-training a source domain classifier on unlabeled data from the target domain improves cross-domain few-shot transfer.		 Cheng Perng Phoo; Bharath Hariharan;
43 Score-Based Generative Modeling Through Stochastic Differential Equations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A general framework for training and sampling from score-based models that unifies and generalizes previous methods, allows likelihood computation, and enables controllable generation.		 Yang Song; Jascha Sohl-Dickstein; Diederik P Kingma; Abhishek Kumar; Stefano Ermon; Ben Poole;
44 Share Or Not? Learning to Schedule Language-Specific Capacity for Multilingual Translation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We investigate and improve parameter-sharing strategies in multilingual Transformers by utilizing conditional computation.		 Biao Zhang; Ankur Bapna; Rico Sennrich; Orhan Firat;
45 Image GANs Meet Differentiable Rendering for Inverse Graphics and Interpretable 3D Neural Rendering
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We marry generative models with differentiable rendering to extract and disentangle 3D knowledge learned implicitly by generative image synthesis models		 Yuxuan Zhang; Wenzheng Chen; Huan Ling; Jun Gao; Yinan Zhang; Antonio Torralba; Sanja Fidler;
46 How Neural Networks Extrapolate: From Feedforward to Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study how neural networks trained by gradient descent extrapolate.		 Keyulu Xu; Mozhi Zhang; Jingling Li; Simon Shaolei Du; Ken-Ichi Kawarabayashi; Stefanie Jegelka;
47 Contrastive Explanations for Reinforcement Learning Via Embedded Self Predictions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduced the embedded self-prediction (ESP) model for producing meaningful and sound contrastive explanations for RL agents.		 Zhengxian Lin; Kin-Ho Lam; Alan Fern;
48 Improved Autoregressive Modeling with Distribution Smoothing
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Inspired by randomized smoothing for adversarial defense, we incorporate randomized smoothing techniques into autoregressive generative modeling.		 Chenlin Meng; Jiaming Song; Yang Song; Shengjia Zhao; Stefano Ermon;
49 MONGOOSE: A Learnable LSH Framework for Efficient Neural Network Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose MONGOOSE, a learnable LSH framework for efficient neural network training.		 Beidi Chen; Zichang Liu; Binghui Peng; Zhaozhuo Xu; Jonathan Lingjie Li; Tri Dao; Zhao Song; Anshumali Shrivastava; Christopher Re;
50 Gradient Projection Memory for Continual Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To avoid catastrophic forgetting in continual learning, we propose a novel approach where a neural network learns new tasks by taking gradient steps in the orthogonal direction to the gradient subspaces deemed important for past tasks.		 Gobinda Saha; Kaushik Roy;
51 Why Are Convolutional Nets More Sample-Efficient Than Fully-Connected Nets?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We construct a single natural distribution on which any fully-connected networks trained with SGD requires \Omega(d^2) samples to generalize while O (1) samples suffices for convolutional architectures.		 Zhiyuan Li; Yi Zhang; Sanjeev Arora;
52 Iterated Learning for Emergent Systematicity in VQA
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use iterated learning to encourage the emergence of structure in the generated programs for neural module networks.		 Ankit Vani; Max Schwarzer; Yuchen Lu; Eeshan Dhekane; Aaron Courville;
53 Coupled Oscillatory Recurrent Neural Network (coRNN): An Accurate and (gradient) Stable Architecture for Learning Long Time Dependencies
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A biologically motivated and discretized ODE based RNN for learning long-term dependencies, with rigorous bounds mitigating the exploding and vanishing gradient problem.		 T. Konstantin Rusch; Siddhartha Mishra;
54 Sparse Quantized Spectral Clustering
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Here, we exploit tools from random matrix theory to make precise statements about how the eigenspectrum of a matrix changes under such nonlinear transformations.		 Zhenyu Liao; Romain Couillet; Michael W. Mahoney;
55 Graph-Based Continual Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose to augment such an array with a learnable random graph that captures pairwise similarities between its samples, and use it not only to learn new tasks but also to guard against forgetting.		 Binh Tang; David S. Matteson;
56 Dynamic Tensor Rematerialization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an online algorithm for rematerialization (recomputing intermediate activations during backpropagation instead of storing them), which enables training under low memory, finding that it is competitive with offline techniques.		 Marisa Kirisame; Steven Lyubomirsky; Altan Haan; Jennifer Brennan; Mike He; Jared Roesch; Tianqi Chen; Zachary Tatlock;
57 Gradient Vaccine: Investigating and Improving Multi-task Optimization in Massively Multilingual Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we attempt to peek into the black-box of multilingual optimization through the lens of loss function geometry.		 Zirui Wang; Yulia Tsvetkov; Orhan Firat; Yuan Cao;
58 CPT: Efficient Deep Neural Network Training Via Cyclic Precision
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose Cyclic Precision Training towards better accuracy-efficiency trade-offs in DNN training.		 Yonggan Fu; Han Guo; Meng Li; Xin Yang; Yining Ding; Vikas Chandra; Yingyan Lin;
59 Learning A Latent Simplex in Input Sparsity Time
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We obtain the first input sparsity runtime algorithm for the problem of learning a latent simplex.		 Ainesh Bakshi; Chiranjib Bhattacharyya; Ravi Kannan; David Woodruff; Samson Zhou;
60 Expressive Power of Invariant and Equivariant Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a theoretical framework able to compare the expressive power of these GNN architectures.		 Waiss Azizian; marc lelarge;
61 Discovering A Set of Policies for The Worst Case Reward
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the problem of how to construct a set of policies that can be composed together to solve a collection of reinforcement learning tasks.		 Tom Zahavy; Andre Barreto; Daniel J Mankowitz; Shaobo Hou; Brendan O’Donoghue; Iurii Kemaev; Satinder Singh;
62 Model-Based Visual Planning with Self-Supervised Functional Distances
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We combine model-based planning with dynamical distance learning to solve visual goal-reaching tasks, using random, unlabeled, experience.		 Stephen Tian; Suraj Nair; Frederik Ebert; Sudeep Dasari; Benjamin Eysenbach; Chelsea Finn; Sergey Levine;
63 Noise Against Noise: Stochastic Label Noise Helps Combat Inherent Label Noise
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: SGD noise induced by stochastic label noise helps escape sharp minima and prevents overconfidence, hence can mitigate the effects of inherent label noise and improve generalization.		 Pengfei Chen; Guangyong Chen; Junjie Ye; jingwei zhao; Pheng-Ann Heng;
64 Contrastive Behavioral Similarity Embeddings for Generalization in Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A contrastive representation learning method which encodes behavioural similarity in RL for improving generalization.		 Rishabh Agarwal; Marlos C. Machado; Pablo Samuel Castro; Marc G Bellemare;
65 VAEBM: A Symbiosis Between Variational Autoencoders and Energy-based Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce an energy-based generative model where the data distribution is defined jointly by a VAE and an energy network.		 Zhisheng Xiao; Karsten Kreis; Jan Kautz; Arash Vahdat;
66 Geometry-Aware Gradient Algorithms for Neural Architecture Search
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Studying the right single-level optimization geometry yields state-of-the-art methods for NAS.		 Liam Li; Mikhail Khodak; Nina Balcan; Ameet Talwalkar;
67 Learning-based Support Estimation in Sublinear Time
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A learning-based algorithm for support size estimation, which given a sufficiently accurate predictor, improves both provably and empirically over state-of-the-art algorithms that do not use a predictor.		 Talya Eden; Piotr Indyk; Shyam Narayanan; Ronitt Rubinfeld; Sandeep Silwal; Tal Wagner;
68 Deciphering and Optimizing Multi-Task Learning: A Random Matrix Approach
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper provides a theoretical analysis of Multi Task Learning schemes for large dimensional data		 Malik Tiomoko; Hafiz Tiomoko Ali; Romain Couillet;
69 Autoregressive Entity Retrieval
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We address entity retrieval by generating their unique name identifiers, left to right, in an autoregressive fashion, and conditioned on the context showing SOTA results in more than 20 datasets with a tiny fraction of the memory of recent systems.		 Nicola De Cao; Gautier Izacard; Sebastian Riedel; Fabio Petroni;
70 Systematic Generalisation with Group Invariant Predictions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Invariance penalties across splits of a biased dataset can improve systematic generalisation		 Faruk Ahmed; Yoshua Bengio; Harm van Seijen; Aaron Courville;
71 Iterative Empirical Game Solving Via Single Policy Best Response
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce two variations of PSRO designed to reduce the amount of simulation required during DRL training.		 Max Smith; Thomas Anthony; Michael Wellman;
72 Understanding The Role of Importance Weighting for Deep Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the theoretical properties of importance weighting for deep learning.		 Da Xu; Yuting Ye; Chuanwei Ruan;
73 Long-tail Learning Via Logit Adjustment
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Adjusting classifier logits based on class priors, either post-hoc or during training, can improve performance on rare classes.		 Aditya Krishna Menon; Sadeep Jayasumana; Ankit Singh Rawat; Himanshu Jain; Andreas Veit; Sanjiv Kumar;
74 DDPNOpt: Differential Dynamic Programming Neural Optimizer
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a new class of optimal-control-theoretic training methods, DDPNOpt, that performs a distinct backward pass inherited with Bellman optimality and generates layer-wise feedback policies to robustify training over existing training methods		 Guan-Horng Liu; Tianrong Chen; Evangelos Theodorou;
75 Learning with Feature Dependent Label Noise: A Progressive Approach
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a progressive label correction approach for noisy label learning task		 Yikai Zhang; Songzhu Zheng; Pengxiang Wu; Mayank Goswami; Chao Chen;
76 Information Laundering for Model Privacy
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose information laundering, a novel framework for enhancing model privacy.		 Xinran Wang; Yu Xiang; Jun Gao; Jie Ding;
77 Mutual Information State Intrinsic Control
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Motivated by the self-consciousness concept in psychology, we propose a new intrinsic objective that encourages the agent to have maximum control on the environment.		 Rui Zhao; Yang Gao; Pieter Abbeel; Volker Tresp; Wei Xu;
78 Benefit of Deep Learning with Non-convex Noisy Gradient Descent: Provable Excess Risk Bound and Superiority to Kernel Methods
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We consider a teacher-student regression model, and eventually show that {\it any} linear estimator can be outperformed by deep learning in a sense of the minimax optimal rate especially for a high dimension setting.		 Taiji Suzuki; Shunta Akiyama;
79 How Does Mixup Help With Robustness and Generalization?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A theoretical point of view for Mixup training		 Linjun Zhang; Zhun Deng; Kenji Kawaguchi; Amirata Ghorbani; James Zou;
80 Dataset Inference: Ownership Resolution in Machine Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce ‘Dataset Inference’ as a new method towards resolving model ownership.		 Pratyush Maini; Mohammad Yaghini; Nicolas Papernot;
81 Individually Fair Gradient Boosting
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an algorithm for training individually fair gradient boosted decision trees classifiers.		 Alexander Vargo; Fan Zhang; Mikhail Yurochkin; Yuekai Sun;
82 Large Scale Image Completion Via Co-Modulated Generative Adversarial Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Bridging the gap between between image-conditional and unconditional GAN architectures via co-modulation		 Shengyu Zhao; Jonathan Cui; Yilun Sheng; Yue Dong; Xiao Liang; Eric I-Chao Chang; Yan Xu;
83 Self-Supervised Policy Adaptation During Deployment
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Generalization across enviroments is known to be hard. We propose a self-supervised method for policy adaptation during deployment that assumes no prior knowledge of the test environment, yet still obtain significant improvements.		 Nicklas Hansen; Rishabh Jangir; Yu Sun; Guillem Aleny�; Pieter Abbeel; Alexei A Efros; Lerrel Pinto; Xiaolong Wang;
84 Sharpness-aware Minimization for Efficiently Improving Generalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Motivated by the connection between geometry of the loss landscape and generalization, we introduce a procedure for simultaneously minimizing loss value and loss sharpness.		 Pierre Foret; Ariel Kleiner; Hossein Mobahi; Behnam Neyshabur;
85 PMI-Masking: Principled Masking of Correlated Spans
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Joint masking of correlated tokens significantly speeds up and improves BERT’s pretraining		 Yoav Levine; Barak Lenz; Opher Lieber; Omri Abend; Kevin Leyton-Brown; Moshe Tennenholtz; Yoav Shoham;
86 Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We argue deeper VAEs should perform better, implement one, and show it outperforms all PixelCNN-based autoregressive models in likelihood, while being substantially more efficient.		 Rewon Child;
87 Data-Efficient Reinforcement Learning with Self-Predictive Representations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a temporal, self-supervised objective for RL agents and show that it significantly improves data efficiency in a setting limited to just 2h of gameplay on Atari.		 Max Schwarzer; Ankesh Anand; Rishab Goel; R Devon Hjelm; Aaron Courville; Philip Bachman;
88 Watch-And-Help: A Challenge for Social Perception and Human-AI Collaboration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce Watch-And-Help (WAH), a challenge for testing social intelligence in agents. For this challenge, we build VirtualHome-Social, a multi-agent household environment, and provide a benchmark including both planning and learning based baselines.		 Xavier Puig; Tianmin Shu; Shuang Li; Zilin Wang; Yuan-Hong Liao; Joshua B. Tenenbaum; Sanja Fidler; Antonio Torralba;
89 A Good Image Generator Is What You Need for High-Resolution Video Synthesis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a framework that leverages contemporary image generators to render high-resolution videos.		 Yu Tian; Jian Ren; Menglei Chai; Kyle Olszewski; Xi Peng; Dimitris N. Metaxas; Sergey Tulyakov;
90 UPDeT: Universal Multi-agent RL Via Policy Decoupling with Transformers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we make the first attempt to explore a universal multi-agent reinforcement learning pipeline, designing one single architecture to fit tasks with the requirement of different observation and action configurations.		 Siyi Hu; Fengda Zhu; Xiaojun Chang; Xiaodan Liang;
91 BUSTLE: Bottom-Up Program Synthesis Through Learning-guided Exploration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use a learned model to guide a bottom-up program synthesis search to efficiently synthesize spreadsheet programs.		 Augustus Odena; Kensen Shi; David Bieber; Rishabh Singh; Charles Sutton; Hanjun Dai;
92 Improving Adversarial Robustness Via Channel-wise Activation Suppressing
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Training with Channel-wise Activation Suppressing (CAS) can help imrove the robustness of adversarial training.		 Yang Bai; Yuyuan Zeng; Yong Jiang; Shu-Tao Xia; Xingjun Ma; Yisen Wang;
93 What Are The Statistical Limits of Offline RL with Linear Function Approximation?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Exponential lower bounds for batch RL with linear function approximation.		 Ruosong Wang; Dean Foster; Sham M. Kakade;
94 Unlearnable Examples: Making Personal Data Unexploitable
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a type of error-minimizing noise that can make training examples unlearnable to deep learning.		 Hanxun Huang; Xingjun Ma; Sarah Monazam Erfani; James Bailey; Yisen Wang;
95 Learning Mesh-Based Simulation with Graph Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a general method for learning the dynamics of complex physics systems accurately and efficiently on meshes		 Tobias Pfaff; Meire Fortunato; Alvaro Sanchez-Gonzalez; Peter Battaglia;
96 Locally Free Weight Sharing for Network Width Search
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: One-shot locally free weight sharing supernet for searching optimal network width		 Xiu Su; Shan You; Tao Huang; Fei Wang; Chen Qian; Changshui Zhang; Chang Xu;
97 Graph Convolution with Low-rank Learnable Local Filters
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Graph convolution model for landmark data with local graph regularization and provable graph signal representation expressiveness and stability.		 Xiuyuan Cheng; Zichen Miao; Qiang Qiu;
98 Regularized Inverse Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose tractable solutions of regularized IRL and algorithms to acquire those solutions for both discrete and continuous control problems.		 Wonseok Jeon; Chen-Yang Su; Paul Barde; Thang Doan; Derek Nowrouzezahrai; Joelle Pineau;
99 Interpreting Graph Neural Networks for NLP With Differentiable Edge Masking
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a novel post-hoc interpretation method for graph neural networks, and apply it to analyse two models from the NLP literature.		 Michael Sejr Schlichtkrull; Nicola De Cao; Ivan Titov;
100 Deep Neural Network Fingerprinting By Conferrable Adversarial Examples
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Proposal of a new property called conferrability for adversarial examples that we use as a method for DNN fingerprinting robust to model extraction.		 Nils Lukas; Yuxuan Zhang; Florian Kerschbaum;
101 Tent: Fully Test-Time Adaptation By Entropy Minimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Deep networks can generalize better during testing by adapting to feedback from their own predictions.		 Dequan Wang; Evan Shelhamer; Shaoteng Liu; Bruno Olshausen; Trevor Darrell;
102 GAN Steerability Without Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Extracting linear & nonlinear semantic directions in GAN latent space without any required optimization		 Nurit Spingarn; Ron Banner; Tomer Michaeli;
103 Contrastive Divergence Learning Is A Time Reversal Adversarial Game
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an alternative derivation of the classical Contrastive divergence method, which reveals that it is in fact an adversarial learning procedure.		 Omer Yair; Tomer Michaeli;
104 Topology-Aware Segmentation Using Discrete Morse Theory
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a new approach to train deep image segmentation networks for better topological accuracy.		 Xiaoling Hu; Yusu Wang; Li Fuxin; Dimitris Samaras; Chao Chen;
105 Are Neural Rankers Still Outperformed By Gradient Boosted Decision Trees?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Abstract: Despite the success of neural models on many major machine learning problems, their effectiveness on traditional Learning-to-Rank (LTR) problems is still not widely acknowledged. …		 Zhen Qin; Le Yan; Honglei Zhuang; Yi Tay; Rama Kumar Pasumarthi; Xuanhui Wang; Michael Bendersky; Marc Najork;
106 Predicting Infectiousness for Proactive Contact Tracing
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Proposes a framework called Proactive Contact Tracing which uses distributed inference of expected Covid-19 infectiousness to provide individualized, private recommendations.		 Yoshua Bengio; Prateek Gupta; Tegan Maharaj; Nasim Rahaman; Martin Weiss; Tristan Deleu; Eilif Benjamin Muller; Meng Qu; victor schmidt; Pierre-luc St-charles; hannah alsdurf; Olexa Bilaniuk; david buckeridge; gaetan caron; pierre luc carrier; Joumana Ghosn; satya ortiz gagne; Christopher Pal; Irina Rish; Bernhard Sch�lkopf; abhinav sharma; Jian Tang; andrew williams;
107 Regularization Matters in Policy Optimization – An Empirical Study on Continuous Control
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that conventional regularization methods (e.g., l2), which have been largely ignored in RL methods, can be very effective in policy optimization on continuous control tasks; we also analyze why they can help from several perspectives.		 Zhuang Liu; Xuanlin Li; Bingyi Kang; Trevor Darrell;
108 Minimum Width for Universal Approximation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We establish the tight bound on width for the universal approximability of neural network.		 Sejun Park; Chulhee Yun; Jaeho Lee; Jinwoo Shin;
109 Towards Robustness Against Natural Language Word Substitutions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Model word substitutions at the vector space using convex hull towards robustness.		 Xinshuai Dong; Hong Liu; Rongrong Ji; Anh Tuan Luu;
110 Dynamics of Deep Equilibrium Linear Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We analyze gradient dynamics of deep equilibrium linear models and mathematically prove its theoretical properties.		 Kenji Kawaguchi;
111 Structured Prediction As Translation Between Augmented Natural Languages
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a unified text-to-text approach to handle a variety of structured prediction tasks in a single model, allowing seamless multi-task training and providing extra benefits on low-resource scenarios.		 Giovanni Paolini; Ben Athiwaratkun; Jason Krone; Jie Ma; Alessandro Achille; RISHITA ANUBHAI; Cicero Nogueira dos Santos; Bing Xiang; Stefano Soatto;
112 How Benign Is Benign Overfitting ?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Interpolating label noise hurts adversarial robustness		 Amartya Sanyal; Puneet K. Dokania; Varun Kanade; Philip Torr;
113 Correcting Experience Replay for Multi-agent Communication
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We improve multi-agent learning by relabelling past experience to better reflect current communication policies		 Sanjeevan Ahilan; Peter Dayan;
114 Emergent Symbols Through Binding in External Memory
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a new architecture, the Emergent Symbol Binding Network, that enables rapid learning of abstract rules and strong generalization of those rules to novel entities.		 Taylor Whittington Webb; Ishan Sinha; Jonathan Cohen;
115 Influence Estimation for Generative Adversarial Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an influence estimation method which predicts how GAN’s output changes if an training instance is abesent, and propose to evaluate harmfulness of the instance by estimating how its absence improves GAN evaluation metric.		 Naoyuki Terashita; Hiroki Ohashi; Yuichi Nonaka; Takashi Kanemaru;
116 PlasticineLab: A Soft-Body Manipulation Benchmark with Differentiable Physics
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a soft-body manipulation benchmark with differentiable physics support.		 Zhiao Huang; Yuanming Hu; Tao Du; Siyuan Zhou; Hao Su; Joshua B. Tenenbaum; Chuang Gan;
117 Implicit Normalizing Flows
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We generalize normalizing flows, allowing the mapping to be implicitly defined by the roots of an equation and enlarging the expressiveness power while retaining the tractability.		 Cheng Lu; Jianfei Chen; Chongxuan Li; Qiuhao Wang; Jun Zhu;
118 Support-set Bottlenecks for Video-text Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use a generative objective to improve the instance discrimination limitations of contrastive learning to set new state-of-the-art results in text-to-video retrieval		 Mandela Patrick; Po-Yao Huang; Yuki Asano; Florian Metze; Alexander G Hauptmann; Joao F. Henriques; Andrea Vedaldi;
119 Winning The L2RPN Challenge: Power Grid Management Via Semi-Markov Afterstate Actor-Critic
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an off-policy actor-critic approach that effectively tackles the unique challenges in power grid management by reinforcement learning, adopting the hierarchical policy together with the afterstate representation.		 Deunsol Yoon; Sunghoon Hong; Byung-Jun Lee; Kee-Eung Kim;
120 Learning Incompressible Fluid Dynamics from Scratch – Towards Fast, Differentiable Fluid Models That Generalize
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an unsupervised training framework for incompressible fluid dynamics that allows neural networks to perform fast, accurate, differentiable fluid simulations and generalize to new domain geometries.		 Nils Wandel; Michael Weinmann; Reinhard Klein;
121 The Traveling Observer Model: Multi-task Learning Through Spatial Variable Embeddings
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Learn a single model across unrelated tasks by embedding their input and output variables in a shared space.		 Elliot Meyerson; Risto Miikkulainen;
122 Grounded Language Learning Fast and Slow
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A language-learning agent with dual-coding external memory meta-learns to combine fast-mapped and semantic lexical knowledge to execute instructions in one-shot..		 Felix Hill; Olivier Tieleman; Tamara von Glehn; Nathaniel Wong; Hamza Merzic; Stephen Clark;
123 Long-tailed Recognition By Routing Diverse Distribution-Aware Experts
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We aim to reduce both the bias and the variance of a long-tailed classifier by RoutIng Diverse Experts (RIDE).		 Xudong Wang; Long Lian; Zhongqi Miao; Ziwei Liu; Stella Yu;
124 Differentially Private Learning Needs Better Features (or Much More Data)
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Linear models with handcrafted features outperform end-to-end CNNs for differentially private learning		 Florian Tramer; Dan Boneh;
125 Unsupervised Object Keypoint Learning Using Local Spatial Predictability
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose PermaKey, a novel method for learning object keypoint representations that leverages local predictability as a measure of objectness.		 Anand Gopalakrishnan; Sjoerd van Steenkiste; J�rgen Schmidhuber;
126 On Statistical Bias In Active Learning: How and When to Fix It
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We formalize the bias introduced by active learning and investigate the situations in which it can be harmful and sometimes even helpful, further introducing novel corrective weights to remove it when doing so is beneficial.		 Sebastian Farquhar; Yarin Gal; Tom Rainforth;
127 Implicit Convex Regularizers of CNN Architectures: Convex Optimization of Two- and Three-Layer Networks in Polynomial Time
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the training problem for various CNN architectures with ReLU activations and introduce equivalent finite dimensional convex formulations that can be used to globally optimize these architectures.		 Tolga Ergen; Mert Pilanci;
128 Generalization in Data-driven Models of Primary Visual Cortex
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a novel network architecture which sets a new state of the art at predicting neural responses to visual input and successfully learns generalizing features of mouse visual cortex (V1).		 Konstantin-Klemens Lurz; Mohammad Bashiri; Konstantin Willeke; Akshay Jagadish; Eric Wang; Edgar Y. Walker; Santiago A Cadena; Taliah Muhammad; Erick Cobos; Andreas S. Tolias; Alexander S Ecker; Fabian H. Sinz;
129 Mathematical Reasoning Via Self-supervised Skip-tree Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We demonstrate that self-supervised language modeling applied to mathematical formulas enables logical reasoning.		 Markus Norman Rabe; Dennis Lee; Kshitij Bansal; Christian Szegedy;
130 Parameterization of Hypercomplex Multiplications
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new parameterization of hypercomplex multiplications for architectural flexibility and effectiveness.		 Aston Zhang; Yi Tay; SHUAI Zhang; Alvin Chan; Anh Tuan Luu; Siu Hui; Jie Fu;
131 Distributional Sliced-Wasserstein and Applications to Generative Modeling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new optimal transport distance based on slicing approach and its applications to generative modeling.		 Khai Nguyen; Nhat Ho; Tung Pham; Hung Bui;
132 Async-RED: A Provably Convergent Asynchronous Block Parallel Stochastic Method Using Deep Denoising Priors
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our work develops a novel deep-regularized asynchronous parallel method with provable convergence guarantees for solving large-scale inverse problems.		 Yu Sun; Jiaming Liu; Yiran Sun; Brendt Wohlberg; Ulugbek Kamilov;
133 DeepAveragers: Offline Reinforcement Learning By Solving Derived Non-Parametric MDPs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The paper introduces and investigates an offline RL approach based on optimally solving a finite-state MDP that is derived from the experience dataset using any latent state representation.		 Aayam Kumar Shrestha; Stefan Lee; Prasad Tadepalli; Alan Fern;
134 Learning from Protein Structure with Geometric Vector Perceptrons
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a novel graph neural network layer to learn from the structure of macromolecules.		 Bowen Jing; Stephan Eismann; Patricia Suriana; Raphael John Lamarre Townshend; Ron Dror;
135 Behavioral Cloning from Noisy Demonstrations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an imitation learning algorithm to learn from non-optimal (noisy) demonstrations without any environment interactions and annotations associated with the demonstrations.		 Fumihiro Sasaki; Ryota Yamashina;
136 Undistillable: Making A Nasty Teacher That CANNOT Teach Students
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the Nasty Teacher, a defensive approach to prevent unauthorized cloning from a teacher model through knowledge distillation.		 Haoyu Ma; Tianlong Chen; Ting-Kuei Hu; Chenyu You; Xiaohui Xie; Zhangyang Wang;
137 Multivariate Probabilistic Time Series Forecasting Via Conditioned Normalizing Flows
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: SOTA Multivariate probabilistic time series forecasting using RNNs or Attention to model the dynamics and normalizing flows for the emission model.		 Kashif Rasul; Abdul-Saboor Sheikh; Ingmar Schuster; Urs M Bergmann; Roland Vollgraf;
138 Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The first successful demonstration that image augmentation can be applied to image-based Deep RL to achieve SOTA performance.		 Denis Yarats; Ilya Kostrikov; Rob Fergus;
139 HW-NAS-Bench: Hardware-Aware Neural Architecture Search Benchmark
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A Hardware-Aware Neural Architecture Search Benchmark		 Chaojian Li; Zhongzhi Yu; Yonggan Fu; Yongan Zhang; Yang Zhao; Haoran You; Qixuan Yu; Yue Wang; Cong Hao; Yingyan Lin;
140 Practical Real Time Recurrent Learning with A Sparse Approximation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show how to make RTRL efficient with sparse RNNs, a sparse approximation, or both.		 Jacob Menick; Erich Elsen; Utku Evci; Simon Osindero; Karen Simonyan; Alex Graves;
141 Random Feature Attention
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a random-feature-based attention that scales linearly in sequence length, and performs on par with strong transformer baselines on language modeling and machine translation.		 Hao Peng; Nikolaos Pappas; Dani Yogatama; Roy Schwartz; Noah Smith; Lingpeng Kong;
142 A Gradient Flow Framework For Analyzing Network Pruning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper establishes the relationship between regularly used importance measures for network pruning and evolution of model parameters under gradient flow, thus providing useful insights into pruning early-on in training.		 Ekdeep Singh Lubana; Robert Dick;
143 Recurrent Independent Mechanisms
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Learning recurrent mechanisms which operate independently, and sparingly interact can lead to better generalization to out of distribution samples.		 Anirudh Goyal; Alex Lamb; Jordan Hoffmann; Shagun Sodhani; Sergey Levine; Yoshua Bengio; Bernhard Sch�lkopf;
144 The Intrinsic Dimension of Images and Its Impact on Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We measure the dimensionality of common used datasets, and experimentally investigate whether the links between dimensionality and learning that have been identified in the manifold learning literature describe the behaviors of deep neural networks.		 Phil Pope; Chen Zhu; Ahmed Abdelkader; Micah Goldblum; Tom Goldstein;
145 Uncertainty Sets for Image Classifiers Using Conformal Prediction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We quantify uncertainty for image classifiers using prediction sets, with detailed experiments on Imagenet Val and V2.		 Anastasios Nikolas Angelopoulos; Stephen Bates; Michael Jordan; Jitendra Malik;
146 Sequential Density Ratio Estimation for Simultaneous Optimization of Speed and Accuracy
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: With a novel sequential density estimation algorithm, we relax critical assumptions of the classical Sequential Probability Ratio Test to be applicable in various real-world scenarios.		 Akinori F Ebihara; Taiki Miyagawa; Kazuyuki Sakurai; Hitoshi Imaoka;
147 Disentangled Recurrent Wasserstein Autoencoder
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the first recurrent Wasserstein Autoencoder for learning disentangled representations of sequential data with theoretical analysis.		 Jun Han; Martin Renqiang Min; Ligong Han; Xuan Zhang; Li Erran Li;
148 Generalization Bounds Via Distillation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper provides a suite of mathematical tools to bound the generalization error of networks which possess low-complexity distillations.		 Daniel Hsu; Ziwei Ji; Matus Telgarsky; Lan Wang;
149 Neural Approximate Sufficient Statistics for Implicit Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We learn low-dimensional near-sufficient statistics by infomax principle to improve likelihood-free inference methods.		 Yanzhi Chen; Dinghuai Zhang; Michael U. Gutmann; Aaron Courville; Zhanxing Zhu;
150 A Panda? No, It’s A Sloth: Slowdown Attacks on Adaptive Multi-Exit Neural Network Inference
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Is the computational savings provided by the input-adaptive ‘multi-exit architectures’ robust against adversarial perturbations? No.		 Sanghyun Hong; Yigitcan Kaya; Ionu?-Vlad Modoranu; Tudor Dumitras;
151 Orthogonalizing Convolutional Layers with The Cayley Transform
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose and evaluate an alternative approach to directly parameterize convolutional layers that are constrained to be orthogonal.		 Asher Trockman; J Zico Kolter;
152 LambdaNetworks: Modeling Long-range Interactions Without Attention
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Scalable framework for capturing long-range interactions between input and structured contextual information, which leads to strong improvements in vision tasks.		 Irwan Bello;
153 Mind The Pad — CNNs Can Develop Blind Spots
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The padding mechanism in CNNs can induce harmful spatial bias in the learned weights and in the feature maps, which can be mitigated with careful architectural choices.		 Bilal Alsallakh; Narine Kokhlikyan; Vivek Miglani; Jun Yuan; Orion Reblitz-Richardson;
154 Meta-GMVAE: Mixture of Gaussian VAE for Unsupervised Meta-Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: we propose a principled unsupervised meta-learning model which meta-learns a set-level variational posterior, by matching it with multi-modal prior distribution obtained by EM.		 Dong Bok Lee; Dongchan Min; Seanie Lee; Sung Ju Hwang;
155 Fast Geometric Projections for Local Robustness Certification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a fast, scalable procedure for checking local robustness in neural networks		 Aymeric Fromherz; Klas Leino; Matt Fredrikson; Bryan Parno; Corina Pasareanu;
156 Fidelity-based Deep Adiabatic Scheduling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new loss for applying supervised deep learning to the problem of scheduling adiabatic quantum computations		 Eli Ovits; Lior Wolf;
157 On Self-Supervised Image Representations for GAN Evaluation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that the state-of-the-art self-supervised representations should be used when comparing GANs on the non-Imagenet datasets		 Stanislav Morozov; Andrey Voynov; Artem Babenko;
158 Retrieval-Augmented Generation for Code Summarization Via Hybrid GNN
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper proposes a novel retrieval-augmented mechanism to augment the code semantics with hybrid graph neural network for source code summarization. To evaluate the proposed approach, we release a new challenging benchmark, crawled from diversified large-scale open-source C projects (total 95k+ unique functions in the dataset).		 Shangqing Liu; Yu Chen; Xiaofei Xie; Jing Kai Siow; Yang Liu;
159 Self-supervised Visual Reinforcement Learning with Object-centric Representations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The combination of object-centric representations and goal-conditioned attention policies helps autonomous agents to learn useful multi-task policies in visual multi-object environments		 Andrii Zadaianchuk; Maximilian Seitzer; Georg Martius;
160 Identifying Nonlinear Dynamical Systems with Multiple Time Scales and Long-range Dependencies
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a novel regularization for ReLU-based vanilla RNN that mitigates the exploding vs. vanishing gradient problem while retaining a simple mathematical structure that makes the RNN’s dynamical systems properties partly analytically tractable		 Dominik Schmidt; Georgia Koppe; Zahra Monfared; Max Beutelspacher; Daniel Durstewitz;
161 Neural Topic Model Via Optimal Transport
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper presents a neural topic model via optimal transport, which can discover more coherent and diverse topics and derive better document representations for both regular and short texts.		 He Zhao; Dinh Phung; Viet Huynh; Trung Le; Wray Buntine;
162 Memory Optimization for Deep Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: MONeT reduces the memory footprint of training while minimizing compute overhead by jointly optimizing checkpointing with operator optimizations.		 Aashaka Shah; Chao-Yuan Wu; Jayashree Mohan; Vijay Chidambaram; Philipp Kraehenbuehl;
163 Stabilized Medical Image Attacks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a stabilized adversarial attack method for medical image analysis.		 Gege Qi; Lijun GONG; Yibing Song; Kai Ma; Yefeng Zheng;
164 Quantifying Differences in Reward Functions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A theoretically principled distance measure on reward functions that is quick to compute and predicts policy training performance.		 Adam Gleave; Michael D Dennis; Shane Legg; Stuart Russell; Jan Leike;
165 MARS: Markov Molecular Sampling for Multi-objective Drug Discovery
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose MARS, a method for multi-objective drug molecule discovery.		 Yutong Xie; Chence Shi; Hao Zhou; Yuwei Yang; Weinan Zhang; Yong Yu; Lei Li;
166 Gauge Equivariant Mesh CNNs: Anisotropic Convolutions on Geometric Graphs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Expressive anisotropic mesh convolution without having to pick arbitrary kernel orientation by using gauge equivariance		 Pim De Haan; Maurice Weiler; Taco Cohen; Max Welling;
167 RMSprop Can Converge with Proper Hyper-parameter
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Towards closing the gap between theory and practice, we prove that RMSprop can converge with proper choice of hyper-parameters under certain conditions.		 Naichen Shi; Dawei Li; Mingyi Hong; Ruoyu Sun;
168 Revisiting Dynamic Convolution Via Matrix Decomposition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Efficient network with dynamic matrix decomposition		 Yunsheng Li; Yinpeng Chen; Xiyang Dai; mengchen liu; Dongdong Chen; Ye Yu; Lu Yuan; Zicheng Liu; Mei Chen; Nuno Vasconcelos;
169 Explainable Deep One-Class Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce an approach to explainable deep anomaly detection based on fully convolutional neural networks.		 Philipp Liznerski; Lukas Ruff; Robert A. Vandermeulen; Billy Joe Franks; Marius Kloft; Klaus Robert Muller;
170 Taking Notes on The Fly Helps Language Pre-Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We improve the efficiency of language pre-training methods through providing better data utilization.		 Qiyu Wu; Chen Xing; Yatao Li; Guolin Ke; Di He; Tie-Yan Liu;
171 Mixed-Features Vectors and Subspace Splitting
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Motivated by metagenomics, recommender systems, dictionary learning, and related problems, this paper introduces subspace splitting(SS): the task of clustering the entries of what we call amixed-features vector, that is, a vector whose subsets of coordinates agree with a collection of subspaces.		 Alejandro Pimentel-Alarc�n; Daniel L. Pimentel-Alarc�n;
172 Neural Pruning Via Growing Regularization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose two new deep network pruning algorithms based a growing regularization paradigm.		 Huan Wang; Can Qin; Yulun Zhang; Yun Fu;
173 Practical Massively Parallel Monte-Carlo Tree Search Applied to Molecular Design
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Novel massively parallel MCTS achieves state-of-the-art score in molecular design benchmark.		 Xiufeng Yang; Tanuj Aasawat; Kazuki Yoshizoe;
174 Empirical Analysis of Unlabeled Entity Problem in Named Entity Recognition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work studys what are the impacts of unlabeled entity problem on NER models and how to effectively eliminate them by a general method.		 Yangming Li; lemao liu; Shuming Shi;
175 Deep Networks and The Multiple Manifold Problem
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove a finite-time generalization result for deep fully-connected neural networks trained by gradient descent to classify structured data, where the required width, depth, and sample complexity depend only on intrinsic properties of the data.		 Sam Buchanan; Dar Gilboa; John Wright;
176 Knowledge Distillation Via Softmax Regression Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We advocate for a method that optimizes the output feature of the penultimate layer of the student network and hence is directly related to representation learning.		 Jing Yang; Brais Martinez; Adrian Bulat; Georgios Tzimiropoulos;
177 Nearest Neighbor Machine Translation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We augment the decoder of a pre-trained machine translation model with a nearest neighbor classifier, substantially improving performance in the single language-pair, multilingual and domain adaptation settings, without any additional training.		 Urvashi Khandelwal; Angela Fan; Dan Jurafsky; Luke Zettlemoyer; Mike Lewis;
178 WrapNet: Neural Net Inference with Ultra-Low-Precision Arithmetic
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We adapt neural networks to integer overflow and extreme low-bit accumulator, and show the efficacy on both software and hardware platforms.		 Renkun Ni; Hong-min Chu; Oscar Castaneda; Ping-yeh Chiang; Christoph Studer; Tom Goldstein;
179 Wandering Within A World: Online Contextualized Few-shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new continual few-shot learning paradigm and a new model. Building upon this setting, we propose a new few-shot learning dataset based on large scale indoor imagery that mimics the visual experience of an agent wandering within a world.		 Mengye Ren; Michael Louis Iuzzolino; Michael Curtis Mozer; Richard Zemel;
180 Few-Shot Learning Via Learning The Representation, Provably
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study when and how much representation learning can help few-shot learning by drastically reducing sample complexity on the target task.		 Simon Shaolei Du; Wei Hu; Sham M. Kakade; Jason D. Lee; Qi Lei;
181 AdaGCN: Adaboosting Graph Convolutional Networks Into Deep Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a novel RNN-like deep graph neural network architecture by incorporating AdaBoost into the computation of network; and the proposed graph convolutional network called AdaGCN~(Adaboosting Graph Convolutional Network) has the ability to efficiently extract knowledge from high-order neighbors of current nodes and then integrates knowledge from different hops of neighbors into the network in an Adaboost way.		 Ke Sun; Zhouchen Lin; Zhanxing Zhu;
182 MultiModalQA: Complex Question Answering Over Text, Tables and Images
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: MultiModalQA: A question answering dataset that requires multi-modal multi-hop reasoning over wikipedia text, tables and images, accompanied by a new multi-hop model for tackling the task.		 Alon Talmor; Ori Yoran; Amnon Catav; Dan Lahav; Yizhong Wang; Akari Asai; Gabriel Ilharco; Hannaneh Hajishirzi; Jonathan Berant;
183 Net-DNF: Effective Deep Modeling of Tabular Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Neural network architecture for tabular data		 Gal Elidan; Liran Katzir; Ran El-Yaniv;
184 Optimal Regularization Can Mitigate Double Descent
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Optimal regularization can provably avoid double-descent in certain settings.		 Preetum Nakkiran; Prayaag Venkat; Sham M. Kakade; Tengyu Ma;
185 Meta Back-Translation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Use meta learning to teach the back-translation model to generate better back-translated sentences.		 Hieu Pham; Xinyi Wang; Yiming Yang; Graham Neubig;
186 Learning A Minimax Optimizer: A Pilot Study
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper introduces the learning to optimize (L2O) methodology, called Twin L2O, for minimax optimization consisting of two LSTMs.		 Jiayi Shen; Xiaohan Chen; Howard Heaton; Tianlong Chen; Jialin Liu; Wotao Yin; Zhangyang Wang;
187 A Wigner-Eckart Theorem for Group Equivariant Convolution Kernels
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We parameterize equivariant convolution kernels by proving a generalization of the Wigner-Eckart theorem for spherical tensor operators.		 Leon Lang; Maurice Weiler;
188 Viewmaker Networks: Learning Views for Unsupervised Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a new generative model that produces views for contrastive learning, matching or outperforming hand-crafted views on image, speech, and wearable sensor datasets		 Alex Tamkin; Mike Wu; Noah Goodman;
189 Scalable Transfer Learning with Expert Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We explore the use of expert representations for transfer with a simple, yet effective, strategy.		 Joan Puigcerver; Carlos Riquelme Ruiz; Basil Mustafa; Cedric Renggli; Andr� Susano Pinto; Sylvain Gelly; Daniel Keysers; Neil Houlsby;
190 Negative Data Augmentation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a framework to do Negative Data Augmentation for generative models and self-supervised learning		 Abhishek Sinha; Kumar Ayush; Jiaming Song; Burak Uzkent; Hongxia Jin; Stefano Ermon;
191 Fantastic Four: Differentiable and Efficient Bounds on Singular Values of Convolution Layers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We derive four provable upper bounds on the largest singular value of convolution layers that are differentiable, independent of size of input image and can be computed efficiently during training with negligible overhead.		 Sahil Singla; Soheil Feizi;
192 CoDA: Contrast-enhanced and Diversity-promoting Data Augmentation for Natural Language Understanding
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a novel data augmentation frame-work dubbed CoDA, which synthesizes diverse and informative augmented examples by integrating multiple transformations organically.		 Yanru Qu; Dinghan Shen; Yelong Shen; Sandra Sajeev; Weizhu Chen; Jiawei Han;
193 Teaching with Commentaries
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a flexible framework for neural network teaching, demonstrate it in various settings, and find that it can improve performance and yield insights about datasets and the training process.		 Aniruddh Raghu; Maithra Raghu; Simon Kornblith; David Duvenaud; Geoffrey Hinton;
194 MixKD: Towards Efficient Distillation of Large-scale Language Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose MixKD, a distillation framework leveraging mixup for large-scale language models.		 Kevin J Liang; Weituo Hao; Dinghan Shen; Yufan Zhou; Weizhu Chen; Changyou Chen; Lawrence Carin;
195 FairFil: Contrastive Neural Debiasing Method for Pretrained Text Encoders
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A debiasing method for large-scale pretrained text encoders via contrastive learning.		 Pengyu Cheng; Weituo Hao; Siyang Yuan; Shijing Si; Lawrence Carin;
196 Probabilistic Numeric Convolutional Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We build a neural network which integrates internal discretization error and missing values probabilistically with GPs		 Marc Anton Finzi; Roberto Bondesan; Max Welling;
197 Computational Separation Between Convolutional and Fully-Connected Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show a computational separation between convolutional and fully-connected networks, proving that the former can leverage strong local structure in the data.		 eran malach; Shai Shalev-Shwartz;
198 On The Stability of Fine-tuning BERT: Misconceptions, Explanations, and Strong Baselines
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We provide an analysis of the fine-tuning instability of BERT-based models and present a simple method to fix it		 Marius Mosbach; Maksym Andriushchenko; Dietrich Klakow;
199 Variational Information Bottleneck for Effective Low-Resource Fine-Tuning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose to use Variational Information Bottleneck to suppress irrelevant features for an effective fine-tuning of large-scale language models in low-resource scenarios.		 Rabeeh Karimi mahabadi; Yonatan Belinkov; James Henderson;
200 Witches’ Brew: Industrial Scale Data Poisoning Via Gradient Matching
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Data poisoning attacks that successfully poison neural networks trained from scratch, even on large-scale datasets like ImageNet.		 Jonas Geiping; Liam H Fowl; W. Ronny Huang; Wojciech Czaja; Gavin Taylor; Michael Moeller; Tom Goldstein;
201 DEBERTA: DECODING-ENHANCED BERT WITH DISENTANGLED ATTENTION
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new model architecture DeBERTa is proposed that improves the BERT and RoBERTa models using disentangled attention and enhanced mask decoder.		 Pengcheng He; Xiaodong Liu; Jianfeng Gao; Weizhu Chen;
202 Optimism in Reinforcement Learning with Generalized Linear Function Approximation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A provably efficient (statistically and computationally) algorithm for reinforcement learning with generalized linear function approximation and no explicit dynamics assumptions.		 Yining Wang; Ruosong Wang; Simon Shaolei Du; Akshay Krishnamurthy;
203 Graph Traversal with Tensor Functionals: A Meta-Algorithm for Scalable Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: GTTF is a meta-algorithm, upon which, many algorithms for graph learning can be implemented, automatically giving them efficiency and scale, yet unbiased learning.		 Elan Sopher Markowitz; Keshav Balasubramanian; Mehrnoosh Mirtaheri; Sami Abu-El-Haija; Bryan Perozzi; Greg Ver Steeg; Aram Galstyan;
204 Diverse Video Generation Using A Gaussian Process Trigger
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Diverse future frame synthesis by modeling the diversity of future states using a Gaussian Process, and using Bayesian inference to sample diverse future states.		 Gaurav Shrivastava; Abhinav Shrivastava;
205 Signatory: Differentiable Computations of The Signature and Logsignature Transforms, on Both CPU and GPU
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Differentiable, GPU-capable implementations of the (log)signature transform via novel algorithms.		 Patrick Kidger; Terry Lyons;
206 MoPro: Webly Supervised Learning with Momentum Prototypes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: MoPro is a new webly-supervised learning framework which advances representation learning using freely-available Web images.		 Junnan Li; Caiming Xiong; Steven Hoi;
207 A Universal Representation Transformer Layer for Few-Shot Image Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: code at: https://github.com/liulu112601/URT		 Lu Liu; William L. Hamilton; Guodong Long; Jing Jiang; Hugo Larochelle;
208 Primal Wasserstein Imitation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new Imitation Learning method based on optimal transport.		 Robert Dadashi; Leonard Hussenot; Matthieu Geist; Olivier Pietquin;
209 Learning Perturbation Sets for Robust Machine Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We learn to characterize real-world changes in well-defined perturbation sets, which allow us train models which are empirically and certifiably robust to real-world adversarial changes.		 Eric Wong; J Zico Kolter;
210 CopulaGNN: Towards Integrating Representational and Correlational Roles of Graphs in Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We distinguish the representational and the correlational information encoded by the graphs in node-level prediction tasks, and propose a novel Copula Graph Neural Network to effectively leverage both information.		 Jiaqi Ma; Bo Chang; Xuefei Zhang; Qiaozhu Mei;
211 On The Critical Role of Conventions in Adaptive Human-AI Collaboration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Training agents that can adapt to new settings in multi-agent games.		 Andy Shih; Arjun Sawhney; Jovana Kondic; Stefano Ermon; Dorsa Sadigh;
212 On The Bottleneck of Graph Neural Networks and Its Practical Implications
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel observation on GNN limitations: in long-range problems, a computational bottleneck causes over-squashing of information.		 Uri Alon; Eran Yahav;
213 The Geometry of Integration in Text Classification RNNs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study text classification RNNs using tools from dynamical systems analysis, finding and explaining the geometry of low-dimensional attractor manifolds.		 Kyle Aitken; Vinay Venkatesh Ramasesh; Ankush Garg; Yuan Cao; David Sussillo; Niru Maheswaranathan;
214 Gradient Descent on Neural Networks Typically Occurs at The Edge of Stability
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We trained neural networks using full-batch gradient descent — you won’t believe what happens next!		 Jeremy Cohen; Simran Kaur; Yuanzhi Li; J Zico Kolter; Ameet Talwalkar;
215 CausalWorld: A Robotic Manipulation Benchmark for Causal Structure and Transfer Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A benchmark to address the challenge of agents transferring their learned skills to related environments; primarily for causal structure and transfer learning.		 Ossama Ahmed; Frederik Tr�uble; Anirudh Goyal; Alexander Neitz; Manuel Wuthrich; Yoshua Bengio; Bernhard Sch�lkopf; Stefan Bauer;
216 Empirical Or Invariant Risk Minimization? A Sample Complexity Perspective
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we provide a sample complexity comparison of the recent invariant risk minimization (IRM) framework with the classic empirical risk minimization (ERM) to answer when is IRM better than ERM in terms of out-of-distribution generalization?		 Kartik Ahuja; Jun Wang; Amit Dhurandhar; Karthikeyan Shanmugam; Kush R. Varshney;
217 Scaling Symbolic Methods Using Gradients for Neural Model Explanation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose a technique for combining gradient-based methods with symbolic techniques to scale such analyses and demonstrate its application for model explanation.		 Subham Sekhar Sahoo; Subhashini Venugopalan; Li Li; Rishabh Singh; Patrick Riley;
218 Control-Aware Representations for Model-based Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Two important questions in this area are how to learn a representation that is amenable to the control problem at hand, and how to achieve an end-to-end framework for representation learning and control. In this paper, we take a few steps towards addressing these questions.		 Brandon Cui; Yinlam Chow; Mohammad Ghavamzadeh;
219 C-Learning: Learning to Achieve Goals Via Recursive Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We reframe the goal-conditioned RL problem as one of predicting and controlling the future state of the world, and derive a principled algorithm to solve this problem.		 Benjamin Eysenbach; Ruslan Salakhutdinov; Sergey Levine;
220 The Bootstrap Framework: Generalization Through The Lens of Online Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show empirical evidence that the performance gap between offline generalization and online optimization is small and propose an alternative framework for studying generalization.		 Preetum Nakkiran; Behnam Neyshabur; Hanie Sedghi;
221 Improving VAEs’ Robustness to Adversarial Attack
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that regularisation methods first developed to obtain ‘disentangled’ VAEs increase the robustness of VAEs to adversarial attack; leveraging this insight we propose an even-more-robust hierarchical VAE.		 Matthew JF Willetts; Alexander Camuto; Tom Rainforth; S Roberts; Christopher C Holmes;
222 Learning Visual Representation from Human Interactions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We learn a self-supervised visual representation from human’s interactions with the visual world. For this study, we collect a dataset of human interactions capturing body part movements and gaze in their daily lives.		 Kiana Ehsani; Daniel Gordon; Thomas Hai Dang Nguyen; Roozbeh Mottaghi; Ali Farhadi;
223 EEC: Learning to Encode and Regenerate Images for Continual Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We train autoencoders with Neural Style Transfer to replay old tasks data for continual learning. The encoded features are converted into centroids and covariances to keep memory footprint from growing while keeping classifier performance stable.		 Ali Ayub; Alan Wagner;
224 Impact of Representation Learning in Linear Bandits
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show representation learning provably improves multi-task linear bandits.		 Jiaqi Yang; Wei Hu; Jason D. Lee; Simon Shaolei Du;
225 MODALS: Modality-agnostic Automated Data Augmentation in The Latent Space
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose an automated data augmentation approach called MODALS (Modality-agnostic Automated Data Augmentation in the Latent Space) to augment data for any modality in a generic way.		 Tsz-Him Cheung; Dit-Yan Yeung;
226 The Recurrent Neural Tangent Kernel
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper we introduce and study the Recurrent Neural Tangent Kernel (RNTK), which provides new insights into the behavior of overparametrized RNNs, including how different time steps are weighted by the RNTK to form the output under different initialization parameters and nonlinearity choices, and how inputs of different lengths are treated.		 Sina Alemohammad; Zichao Wang; Randall Balestriero; Richard Baraniuk;
227 Projected Latent Markov Chain Monte Carlo: Conditional Sampling of Normalizing Flows
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce and demonstrate a novel MCMC technique for sampling from the exact conditional distributions known by normalizing flows.		 Chris Cannella; Mohammadreza Soltani; Vahid Tarokh;
228 Learning The Pareto Front with Hypernetworks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel approach for learning the entire Pareto front using hypernetworks		 Aviv Navon; Aviv Shamsian; Gal Chechik; Ethan Fetaya;
229 Estimating and Evaluating Regression Predictive Uncertainty in Deep Object Detectors
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we focus on estimating predictive distributions for bounding box regression output with variance networks.		 Ali Harakeh; Steven L. Waslander;
230 Predicting Classification Accuracy When Adding New Unobserved Classes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new prediction method of multiclass classification accuracy for an increased number of classes.		 Yuli Slavutsky; Yuval Benjamini;
231 BRECQ: Pushing The Limit of Post-Training Quantization By Block Reconstruction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose a novel PTQ framework, dubbed BRECQ, which pushes the limits of bitwidth in PTQ down to INT2 for the first time.		 Yuhang Li; Ruihao Gong; Xu Tan; Yang Yang; Peng Hu; Qi Zhang; Fengwei Yu; Wei Wang; Shi Gu;
232 No MCMC for Me: Amortized Sampling for Fast and Stable Training of Energy-based Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a new generator-based approach for training EBMs and demonstrate that it trains models which obtain high likelihood and overcomes stability issues common in EBM training.		 Will Sussman Grathwohl; Jacob Jin Kelly; Milad Hashemi; Mohammad Norouzi; Kevin Swersky; David Duvenaud;
233 GraphCodeBERT: Pre-training Code Representations with Data Flow
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present GraphCodeBERT, a pre-trained model for programming language that considers the inherent structure of code.		 Daya Guo; Shuo Ren; Shuai Lu; Zhangyin Feng; Duyu Tang; Shujie LIU; Long Zhou; Nan Duan; Alexey Svyatkovskiy; Shengyu Fu; Michele Tufano; Shao Kun Deng; Colin Clement; Dawn Drain; Neel Sundaresan; Jian Yin; Daxin Jiang; Ming Zhou;
234 Conservative Safety Critics for Exploration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Safe exploration in reinforcement learning can be achieved by constraining policy learning with conservative safety estimates of the environment.		 Homanga Bharadhwaj; Aviral Kumar; Nicholas Rhinehart; Sergey Levine; Florian Shkurti; Animesh Garg;
235 Improve Object Detection with Feature-based Knowledge Distillation: Towards Accurate and Efficient Detectors
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose two knowledge distillation methods on object detection – attention-guided distillation and non-local distillation which lead to 4.1 AP improvements on Faster RCNN101 in MS COCO2017.		 Linfeng Zhang; Kaisheng Ma;
236 A Temporal Kernel Approach for Deep Learning with Continuous-time Information
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a temporal kernel learning approach based on random features and reparameterization to characterize the continuous-time information in deep learning models.		 Da Xu; Chuanwei Ruan; Evren Korpeoglu; Sushant Kumar; Kannan Achan;
237 For Self-supervised Learning, Rationality Implies Generalization, Provably
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove and empirically demonstrate generalization bounds for algorithms that fit a simple classifier on a representation learned via self-supervision; we obtain non-vacuous bounds for such top-performing algorithms on both CIFAR-10 and ImageNet.		 Yamini Bansal; Gal Kaplun; Boaz Barak;
238 How to Find Your Friendly Neighborhood: Graph Attention Design with Self-Supervision
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a method that self-supervise graph attention through edges and it should be designed according to the average degree and homophily of graphs.		 Dongkwan Kim; Alice Oh;
239 Interpretable Models for Granger Causality Using Self-explaining Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an interpretable framework for inferring Granger causality based on self-explaining neural networks.		 Ricards Marcinkevics; Julia E Vogt;
240 Meta-learning Symmetries By Reparameterization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A method for automatically meta-learning and encoding equivariances into neural networks.		 Allan Zhou; Tom Knowles; Chelsea Finn;
241 Removing Undesirable Feature Contributions Using Out-of-Distribution Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a simple method, Out-of-distribution data Augmented Training (OAT), to leverage OOD data for adversarial and standard learning.		 Saehyung Lee; Changhwa Park; Hyungyu Lee; Jihun Yi; Jonghyun Lee; Sungroh Yoon;
242 Mind The Gap When Conditioning Amortised Inference in Sequential Latent-Variable Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show how a common model assumption in amortised variational inference with sequential LVMS leads to a suboptimality and how to prevent it.		 Justin Bayer; Maximilian Soelch; Atanas Mirchev; Baris Kayalibay; Patrick van der Smagt;
243 On The Universality of The Double Descent Peak in Ridgeless Regression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove a distribution-independent lower bound for the generalization error of ridgeless (random) features regression under weak assumptions, showing universal sensitivity to label noise around the interpolation threshold.		 David Holzm�ller;
244 Fair Mixup: Fairness Via Interpolation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To improve the generalizability of fair classifiers, we propose fair mixup, a new data augmentation strategy for imposing the fairness constraint.		 Ching-Yao Chuang; Youssef Mroueh;
245 Self-supervised Learning from A Multi-view Perspective
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: From a multi-view learning perspective, this paper provides both theoretical and empirical analysis on self-supervised learning.		 Yao-Hung Hubert Tsai; Yue Wu; Ruslan Salakhutdinov; Louis-Philippe Morency;
246 Integrating Categorical Semantics Into Unsupervised Domain Translation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a method for learning domain invariant categorial semantics which enable UDT on two setups.		 Samuel Lavoie-Marchildon; Faruk Ahmed; Aaron Courville;
247 The Unreasonable Effectiveness of Patches in Deep Convolutional Kernels Methods
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Patch-based representation is a key ingredient for competitive convolutional kernel methods.		 Louis THIRY; Michael Arbel; Eugene Belilovsky; Edouard Oyallon;
248 Open Question Answering Over Tables and Text
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the new task of answering open-domain questions answering over web tables and text and design new techniques: 1) fused retrieval 2) cross-block reader to resolve the challenges posed in the new task.		 Wenhu Chen; Ming-Wei Chang; Eva Schlinger; William Yang Wang; William W. Cohen;
249 Evaluation of Similarity-based Explanations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We investigated empirically which of the relevance metrics (e.g. similarity of hidden layer, influence function, etc.) are appropriate for similarity-based explanation.		 Kazuaki Hanawa; Sho Yokoi; Satoshi Hara; Kentaro Inui;
250 A Diffusion Theory For Deep Learning Dynamics: Stochastic Gradient Descent Exponentially Favors Flat Minima
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove that, benefited from the Hessian-dependent covariance of stochastic gradient noise, SGD favors flat minima exponentially more than sharp minima.		 Zeke Xie; Issei Sato; Masashi Sugiyama;
251 How Much Over-parameterization Is Sufficient to Learn Deep ReLU Networks?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We establish learning guarantees for deep ReLU networks with width polylogarithmic in sample size and the inverse of the target error.		 Zixiang Chen; Yuan Cao; Difan Zou; Quanquan Gu;
252 Auction Learning As A Two-Player Game
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We formulate the auction learning problem as a two player game with a stationary utility functions and explore the advantages of such an approach.		 Jad Rahme; Samy Jelassi; S. Matthew Weinberg;
253 Robust Reinforcement Learning on State Observations with Learned Optimal Adversary
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the robustness of RL agents under perturbations on states and find that using an optimal adversary learned online in an alternating training manner can improve the robustness of agent policy.		 Huan Zhang; Hongge Chen; Duane S Boning; Cho-Jui Hsieh;
254 Optimizing Memory Placement Using Evolutionary Graph Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We combine evolutionary and gradient-based reinforcement learning to tackle the large search spaces needed to map tensors to memory, yielding up to 78% speedup on BERT and ResNet on a deep learning inference chip.		 Shauharda Khadka; Estelle Aflalo; Mattias Mardar; Avrech Ben-David; Santiago Miret; Shie Mannor; Tamir Hazan; Hanlin Tang; Somdeb Majumdar;
255 Hierarchical Autoregressive Modeling for Neural Video Compression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Specifically, we view recent neural video compression methods (Lu et al., 2019; Yang et al., 2020b; Agustssonet al., 2020) as instances of a generalized stochastic temporal autoregressive trans-form, and propose avenues for enhancement based on this insight.		 Ruihan Yang; Yibo Yang; Joseph Marino; Stephan Mandt;
256 Individually Fair Rankings
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an algorithm for training individually fair learning-to-rank systems using optimal transport tools.		 Amanda Bower; Hamid Eftekhari; Mikhail Yurochkin; Yuekai Sun;
257 Learning Neural Generative Dynamics for Molecular Conformation Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel probabilistic framework to generate valid and diverse molecular conformations. Reaching state-of-the-art results on conformation generation and inter-atomic distance modeling.		 Minkai Xu; Shitong Luo; Yoshua Bengio; Jian Peng; Jian Tang;
258 Efficient Certified Defenses Against Patch Attacks on Image Classifiers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a method for certifying robustness against adversarial patches that combines high certified accuracy with efficient inference while maintaining strong performance on clean data.		 Jan Hendrik Metzen; Maksym Yatsura;
259 Convex Regularization Behind Neural Reconstruction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work proposes a finite-dimensional convex dual of a two-layer fully-convolutional ReLU network for denoising problems, and uses it for interpretation of neural network training and predictions.		 Arda Sahiner; Morteza Mardani; Batu Ozturkler; Mert Pilanci; John M. Pauly;
260 Targeted Attack Against Deep Neural Networks Via Flipping Limited Weight Bits
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a targeted attack method against the deployed DNN via flipping a few binary weight bits.		 Jiawang Bai; Baoyuan Wu; Yong Zhang; Yiming Li; Zhifeng Li; Shu-Tao Xia;
261 Generalized Multimodal ELBO
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a generalized ELBO for modeling multiple data types in a scalable and self-supervised way.		 Thomas Marco Sutter; Imant Daunhawer; Julia E Vogt;
262 Large-width Functional Asymptotics for Deep Gaussian Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We establish the convergence of infinitely wide feed-forward deep neural networks in function space.		 Daniele Bracale; Stefano Favaro; Sandra Fortini; Stefano Peluchetti;
263 Distributed Momentum for Byzantine-resilient Stochastic Gradient Descent
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An inexpensive method to substantially improve the effectiveness of existing Byzantine-resilient SGD defenses, assessed against state-of-the-art attacks and supported by theoretical insights.		 El Mahdi El Mhamdi; Rachid Guerraoui; S�bastien Rouault;
264 Fully Unsupervised Diversity Denoising with Convolutional Variational Autoencoders
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: DivNoising performs fully unsupervised diversity denoising using fully convolutional variational autoencoders and achieves SOTA results for a number of well known datasets while also enabling VAE-like sampling		 Mangal Prakash; Alexander Krull; Florian Jug;
265 Auxiliary Learning By Implicit Differentiation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Learn to combine auxiliary tasks in a nonlinear fashion and to design them automatically.		 Aviv Navon; Idan Achituve; Haggai Maron; Gal Chechik; Ethan Fetaya;
266 Balancing Constraints and Rewards with Meta-Gradient D4PG
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper uses meta-gradients to perform soft-constrained Reinforcement Learning (RL) optimization		 Dan A. Calian; Daniel J Mankowitz; Tom Zahavy; Zhongwen Xu; Junhyuk Oh; Nir Levine; Timothy Mann;
267 Adversarially Guided Actor-Critic
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a third protagonist, the adversary. While this adversary mimics the actor by minimizing the KL-divergence between their respective action distributions, the actor maximizes the log-probability difference between its action and that of the adversary in combination with maximizing expected rewards.		 Yannis Flet-Berliac; Johan Ferret; Olivier Pietquin; Philippe Preux; Matthieu Geist;
268 DARTS-: Robustly Stepping Out of Performance Collapse Without Indicators
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Indicator-free approach to stabilize DARTS		 Xiangxiang Chu; Xiaoxing Wang; Bo Zhang; Shun Lu; Xiaolin Wei; Junchi Yan;
269 Are Wider Nets Better Given The Same Number of Parameters?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that increasing network width leads to better performance even when the number of weights remains fixed.		 Anna Golubeva; Guy Gur-Ari; Behnam Neyshabur;
270 Optimal Conversion of Conventional Artificial Neural Networks to Spiking Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose and validate an optimal pipeline that efficiently converts conventional artificial neural networks to spiking neural networks with almost no accuracy loss in a fairly short simulation length.		 Shikuang Deng; Shi Gu;
271 Deep Equals Shallow for ReLU Networks in Kernel Regimes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that for ReLU activations, the kernels derived from deep fully-connected networks have essentially the same approximation properties as their shallow two-layer counterpart, namely the same eigenvalue decay for the corresponding integral operator.		 Alberto Bietti; Francis Bach;
272 Graph Coarsening with Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We significantly improve the quality of existing graph coarsening algorithms with graph neural network.		 Chen Cai; Dingkang Wang; Yusu Wang;
273 Early Stopping in Deep Networks: Double Descent and How to Eliminate It
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Epoch wise double descent can be explained as a superposition of two or more bias-variance tradeoffs that arise because different parts of the network are learned at different epochs.		 Reinhard Heckel; Fatih Furkan Yilmaz;
274 Efficient Inference of Flexible Interaction in Spiking-neuron Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An efficient conjugate EM algorithm for nonlinear multivariate Hawkes processes based on auxiliary latent variables augmentation.		 Feng Zhou; Yixuan Zhang; Jun Zhu;
275 DICE: Diversity in Deep Ensembles Via Conditional Redundancy Adversarial Estimation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Driven by arguments from information theory, we introduce a new learning strategy for deep ensembles that increases diversity among members: we adversarially prevent features from being conditionally redundant, i.e., predictable from each other.		 Alexandre Rame; Matthieu Cord;
276 Inductive Representation Learning in Temporal Networks Via Causal Anonymous Walks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The paper proposes Causal Anonymous Walks (CAW) as an effective way to encode the dynamic laws that govern the evolution of temporal networks, which significantly improves inductive representation learning on those networks.		 Yanbang Wang; Yen-Yu Chang; Yunyu Liu; Jure Leskovec; Pan Li;
277 FairBatch: Batch Selection for Model Fairness
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We address model fairness via the lens of bilevel optimization and propose a batch selection algorithm called FairBatch, which is easy to adopt, has state-of-the-art performance, and is compatible with existing batch selection techniques.		 Yuji Roh; Kangwook Lee; Steven Euijong Whang; Changho Suh;
278 Representation Balancing Offline Model-based Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present RepB-SDE, a framework for balancing the model representation with stationary distribution estimation, aiming at obtaining a model robust to the distribution shift that arises in off-policy and offline RL.		 Byung-Jun Lee; Jongmin Lee; Kee-Eung Kim;
279 Accelerating Convergence of Replica Exchange Stochastic Gradient MCMC Via Variance Reduction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a variance-reduced replica-exchange stochastic gradient Langevin dynamics to reduce the variance of the energy estimators to accelerate the convergence.		 Wei Deng; Qi Feng; Georgios P. Karagiannis; Guang Lin; Faming Liang;
280 The Importance of Pessimism in Fixed-Dataset Policy Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A unified conceptual and mathematical framework for fixed-dataset policy optimization algorithms, revealing the importance of uncertainty and pessimism.		 Jacob Buckman; Carles Gelada; Marc G Bellemare;
281 Interpreting Knowledge Graph Relation Representation from Word Embeddings
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Interpreting the structure of knowledge graph relation representation using insight from word embeddings.		 Carl Allen; Ivana Balazevic; Timothy Hospedales;
282 Hopfield Networks Is All You Need
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel continuous Hopfield network is proposed whose update rule is the attention mechanism of the transformer model and which can be integrated into deep learning architectures.		 Hubert Ramsauer; Bernhard Sch�fl; Johannes Lehner; Philipp Seidl; Michael Widrich; Lukas Gruber; Markus Holzleitner; Thomas Adler; David Kreil; Michael K Kopp; G�nter Klambauer; Johannes Brandstetter; Sepp Hochreiter;
283 Uncertainty Estimation and Calibration with Finite-State Probabilistic RNNs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A method to estimate and calibrate uncertainty in recurrent state transitions.		 Cheng Wang; Carolin Lawrence; Mathias Niepert;
284 Understanding The Failure Modes of Out-of-distribution Generalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this theoretical study, we explain why machine learning models rely on spuriously correlated features in the dataset and fail at out-of-distribution generalization.		 Vaishnavh Nagarajan; Anders Andreassen; Behnam Neyshabur;
285 Generative Language-Grounded Policy in Vision-and-Language Navigation with Bayes’ Rule
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the novel generative language-grounded policy for vision-and-language navigation(VLN).		 Shuhei Kurita; Kyunghyun Cho;
286 Emergent Road Rules In Multi-Agent Driving Environments
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In multi-agent driving environments with noisy perception, driving conventions emerge		 Avik Pal; Jonah Philion; Yuan-Hong Liao; Sanja Fidler;
287 Wasserstein-2 Generative Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a new end-to-end algorithm to compute optimal transport maps between continuous distributions without introducing bias or resorting to minimax optimization.		 Alexander Korotin; Vage Egiazarian; Arip Asadulaev; Alexander Safin; Evgeny Burnaev;
288 Vulnerability-Aware Poisoning Mechanism for Online RL with Unknown Dynamics
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the first poisoning algorithm against deep policy-based RL methods, without any prior knowledge of the environment, covering heterogeneous poisoning models.		 Yanchao Sun; Da Huo; Furong Huang;
289 Tomographic Auto-Encoder: Unsupervised Bayesian Recovery of Corrupted Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Recovering accurate posterior distributions for unsupervised data recovery		 Francesco Tonolini; Andreas Damianou; Pablo Garcia Moreno; Roderick Murray-Smith;
290 Monotonic Kronecker-Factored Lattice
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show how to effectively and efficiently learn flexible and interpretable monotonic functions using Kronecker-Factored Lattice, an efficient reparameterization of flexible monotonic lattice regression via Kronecker product.		 William Taylor Bakst; Nobuyuki Morioka; Erez Louidor;
291 LEAF: A Learnable Frontend for Audio Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a lightweight learnable frontend of audio that can replace fixed features over a wide range of audio classification tasks.		 Neil Zeghidour; Olivier Teboul; F�lix de Chaumont Quitry; Marco Tagliasacchi;
292 Federated Learning Via Posterior Averaging: A New Perspective and Practical Algorithms
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new approach to federated learning that generalizes federated optimization, combines local MCMC-based sampling with global optimization-based posterior inference, and achieves competitive results on challenging benchmarks.		 Maruan Al-Shedivat; Jennifer Gillenwater; Eric Xing; Afshin Rostamizadeh;
293 Rank The Episodes: A Simple Approach for Exploration in Procedurally-Generated Environments
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Encouraging exploration via ranking the past episodes and reproducing past good exploration behaviors with imitation learning.		 Daochen Zha; Wenye Ma; Lei Yuan; Xia Hu; Ji Liu;
294 Partitioned Learned Bloom Filters
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Here we show how to frame the problem of optimal model utilization as an optimization problem, and using our framework derive algorithms that can achieve near-optimal performance in many cases.		 Kapil Vaidya; Eric Knorr; Michael Mitzenmacher; Tim Kraska;
295 Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper improves the learning of dense text retrieval using ANCE, which selects global negatives with bigger gradient norms using an asynchronously updated ANN index.		 Lee Xiong; Chenyan Xiong; Ye Li; Kwok-Fung Tang; Jialin Liu; Paul N. Bennett; Junaid Ahmed; Arnold Overwikj;
296 AUXILIARY TASK UPDATE DECOMPOSITION: THE GOOD, THE BAD AND THE NEUTRAL
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We improve how we use auxiliary task data for model pre-training by decomposing gradient updates into components guided by the primary task		 Lucio M. Dery; Yann Dauphin; David Grangier;
297 SSD: A Unified Framework for Self-Supervised Outlier Detection
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose SSD, an outlier detector based on only unlabeled training data.		 Vikash Sehwag; Mung Chiang; Prateek Mittal;
298 Ask Your Humans: Using Human Instructions to Improve Generalization in Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To facilitate the automatic decomposition of hierarchical tasks, we propose the use of step-by-step human demonstrations in the form of natural language instructions and action trajectories. We introduce a dataset of such demonstrations in a crafting-based grid world.		 Valerie Chen; Abhinav Gupta; Kenneth Marino;
299 Revisiting Few-sample BERT Fine-tuning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper is a study of fine-tuning of BERT contextual representations, with focus on commonly observed instabilities in few-sample scenarios.		 Tianyi Zhang; Felix Wu; Arzoo Katiyar; Kilian Q Weinberger; Yoav Artzi;
300 Tilted Empirical Risk Minimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that tilted empirical risk minimization (TERM) can be used for enforcing fairness between subgroups, mitigating the effect of outliers, and handling class imbalance, all in a unified framework.		 Tian Li; Ahmad Beirami; Maziar Sanjabi; Virginia Smith;
301 Deep Neural Tangent Kernel and Laplace Kernel Have The Same RKHS
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove that the reproducing kernel Hilbert spaces of a deep neural tangent kernel and the Laplace kernel include the same set of functions.		 Lin Chen; Sheng Xu;
302 On The Transfer of Disentangled Representations in Realistic Settings
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We scale disentangled representation learning to a new realistic dataset and conduct a large-scale empirical study on OOD generalization. We introduce a new high-resolution dataset with over 1M simulated images and 1k annotated real-world images of the same setup.		 Andrea Dittadi; Frederik Tr�uble; Francesco Locatello; Manuel Wuthrich; Vaibhav Agrawal; Ole Winther; Stefan Bauer; Bernhard Sch�lkopf;
303 Calibration Tests Beyond Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Unifying framework for calibration evaluations and tests of probabilistic predictive models		 David Widmann; Fredrik Lindsten; Dave Zachariah;
304 Overparameterisation and Worst-case Generalisation: Friend Or Foe?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Overparameterised models’ worst-subgroup performance can be improved via post-hoc processing.		 Aditya Krishna Menon; Ankit Singh Rawat; Sanjiv Kumar;
305 You Only Need Adversarial Supervision for Semantic Image Synthesis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose OASIS, a novel model for multi-modal semantic image synthesis, improving over previous methods in terms of image quality and diversity while only using adversarial supervision.		 Edgar Sch�nfeld; Vadim Sushko; Dan Zhang; Juergen Gall; Bernt Schiele; Anna Khoreva;
306 Learning to Recombine and Resample Data For Compositional Generalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper investigates a data augmentation procedure based on two weaker principles: recombination and resampling, and finds that it is sufficient to induce many of the compositional generalizations studied in previous work.		 Ekin Aky�rek; Afra Feyza Aky�rek; Jacob Andreas;
307 A Critique of Self-Expressive Deep Subspace Clustering
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Here we show theoretically and experimentally that there are a number of flaws with many existing self-expressive deep subspace clustering models.		 Benjamin David Haeffele; Chong You; Rene Vidal;
308 INT: An Inequality Benchmark for Evaluating Generalization in Theorem Proving
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce INT, a synthetic INequality Theorem proving benchmark, to tackle the data sparsity and out-of-distribution problems for theorem proving and benchmarked transformer-based and GNN-based agents’ generalization performance.		 Yuhuai Wu; Albert Jiang; Jimmy Ba; Roger Baker Grosse;
309 Improved Estimation of Concentration Under Lp-Norm Distance Metrics Using Half Spaces
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that concentration of measure does not prohibit the existence of adversarially robust classifiers using a novel method of empirical concentration estimation.		 Jack Prescott; Xiao Zhang; David Evans;
310 Adaptive Federated Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose adaptive federated optimization techniques, and highlight their improved performance over popular methods such as FedAvg.		 Sashank J. Reddi; Zachary Charles; Manzil Zaheer; Zachary Garrett; Keith Rush; Jakub Konecn�; Sanjiv Kumar; Hugh Brendan McMahan;
311 On The Dynamics of Training Attention Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we set up a simple text classification task and study the dynamics of training a simple attention-based classification model using gradient descent.		 Haoye Lu; Yongyi Mao; Amiya Nayak;
312 Linear Convergent Decentralized Optimization with Compression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A Linear Convergent Decentralized Optimization with Communication Compression		 Xiaorui Liu; Yao Li; Rongrong Wang; Jiliang Tang; Ming Yan;
313 Efficient Transformers in Reinforcement Learning Using Actor-Learner Distillation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Actor-Learner Distillation uses an continual form of distillation to attain the high sample-efficiency of transformers while maintaining the reduced total training time of LSTMs in RL applications.		 Emilio Parisotto; Russ Salakhutdinov;
314 Large Associative Memory Problem in Neurobiology and Machine Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our paper proposes a microscopic biologically-plausible theory of modern Hopfield networks.		 Dmitry Krotov; John J. Hopfield;
315 Protecting DNNs from Theft Using An Ensemble of Diverse Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Discontinuous predictions produced by an ensemble of diverse models can be used to create an effective defense against model stealing attacks.		 Sanjay Kariyappa; Atul Prakash; Moinuddin K Qureshi;
316 Proximal Gradient Descent-Ascent: Variable Convergence Under KL Geometry
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Specifically, we show that proximal-GDA admits a novel Lyapunov function, which monotonically decreases in the minimax optimization process and drives the variable sequences to a critical point.		 Ziyi Chen; Yi Zhou; Tengyu Xu; Yingbin Liang;
317 Contextual Dropout: An Efficient Sample-Dependent Dropout Module
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose contextual dropout as a scalable sample-dependent dropout method, which makes the dropout probabilities depend on the input covariates of each data sample.		 XINJIE FAN; Shujian Zhang; Korawat Tanwisuth; Xiaoning Qian; Mingyuan Zhou;
318 MIROSTAT: A NEURAL TEXT DECODING ALGORITHM THAT DIRECTLY CONTROLS PERPLEXITY
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We provide a new text decoding algorithm that directly controls perplexity and hence several important attributes of generated text.		 Sourya Basu; Govardana Sachitanandam Ramachandran; Nitish Shirish Keskar; Lav R. Varshney;
319 DialoGraph: Incorporating Interpretable Strategy-Graph Networks Into Negotiation Dialogues
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose DialoGraph, a negotiation dialogue system that leverages Graph Attention Networks to model complex negotiation strategies while providing interpretability for the model via intermediate structures.		 Rishabh Joshi; Vidhisha Balachandran; Shikhar Vashishth; Alan Black; Yulia Tsvetkov;
320 Multi-Time Attention Networks for Irregularly Sampled Time Series
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper presents a new deep learning architecture for learning with sparse and irregularly sampled multivariate time series.		 Satya Narayan Shukla; Benjamin Marlin;
321 Learning Energy-Based Generative Models Via Coarse-to-Fine Expanding and Sampling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a coarse-to-fine energy-based model learning scheme for generative modeling.		 Yang Zhao; Jianwen Xie; Ping Li;
322 Unsupervised Audiovisual Synthesis Via Exemplar Autoencoders
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an unsupervised approach that converts the input speech of any individual into audiovisual streams of potentially-infinitely many output speakers.		 Kangle Deng; Aayush Bansal; Deva Ramanan;
323 A Learning Theoretic Perspective on Local Explainability
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we explore connections between interpretable machine learning and learning theory through the lens of local approximation explanations.		 Jeffrey Li; Vaishnavh Nagarajan; Gregory Plumb; Ameet Talwalkar;
324 SEED: Self-supervised Distillation For Visual Representation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose SEED, a self-supervised distillation technique for visual representation learning.		 Zhiyuan Fang; Jianfeng Wang; Lijuan Wang; Lei Zhang; Yezhou Yang; Zicheng Liu;
325 Isometric Propagation Network for Generalized Zero-shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We improve the current zero-shot learning performance by a dynamic alignment between the semantic space and visual space that encourages the isometry of the class-prototype propagation procedures in the two spaces. To evaluate the generalization capability of IPN, we further build two larger benchmarks with more diverse unseen classes, and demonstrate the advantages of IPN on them.		 Lu Liu; Tianyi Zhou; Guodong Long; Jing Jiang; Xuanyi Dong; Chengqi Zhang;
326 Effective and Efficient Vote Attack on Capsule Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an effective and efficient vote attack to create adversarial examples and bypass adversarial example detection on Capsule Networks.		 Jindong Gu; Baoyuan Wu; Volker Tresp;
327 Heteroskedastic and Imbalanced Deep Learning with Adaptive Regularization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a data-dependent regularization technique for learning heteroskedastic and imbalanced datasets.		 Kaidi Cao; Yining Chen; Junwei Lu; Nikos Arechiga; Adrien Gaidon; Tengyu Ma;
328 Continuous Wasserstein-2 Barycenter Estimation Without Minimax Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a new algorithm to compute Wasserstein-2 barycenters of continuous distributions powered by a straightforward optimization procedure without introducing bias or a generative model.		 Alexander Korotin; Lingxiao Li; Justin Solomon; Evgeny Burnaev;
329 Neural Thompson Sampling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a new algorithm, called Neural Thompson Sampling, which adapts deep neural networks for both exploration and exploitation.		 Weitong ZHANG; Dongruo Zhou; Lihong Li; Quanquan Gu;
330 Symmetry, Conservation Laws, and Learning Dynamics in Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: By exploiting architectural symmetry, our work demonstrates that we can analytically describe the learning dynamics of various parameter combinations at finite learning rates and batch sizes for state of the art architectures trained on any dataset.		 Daniel Kunin; Javier Sagastuy-Brena; Surya Ganguli; Daniel LK Yamins; Hidenori Tanaka;
331 Neural Gradients Are Near-lognormal: Improved Quantized and Sparse Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Considering this, we suggest two closed-form analytical methods to reduce the computational and memory burdens of neural gradients.		 Brian Chmiel; Liad Ben-Uri; Moran Shkolnik; Elad Hoffer; Ron Banner; Daniel Soudry;
332 RODE: Learning Roles to Decompose Multi-Agent Tasks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a scalable role-based multi-agent learning method which effectively discovers roles based on joint action space decomposition according to action effects, establishing a new state of the art on the StarCraft multi-agent benchmark.		 Tonghan Wang; Tarun Gupta; Anuj Mahajan; Bei Peng; Shimon Whiteson; Chongjie Zhang;
333 Revisiting Hierarchical Approach for Persistent Long-Term Video Prediction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a simple yet effective hierarchical video prediction model that can synthesize future frames orders of magnitude longer than existing methods (thousands frames)		 Wonkwang Lee; Whie Jung; Han Zhang; Ting Chen; Jing Yu Koh; Thomas Huang; Hyungsuk Yoon; Honglak Lee; Seunghoon Hong;
334 Physics-aware, Probabilistic Model Order Reduction with Guaranteed Stability
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel physics-aware, generative, probabilistic state-space model for learning an effective, lower-dimensional description that can produce long-term predictions.		 Sebastian Kaltenbach; Phaedon Stelios Koutsourelakis;
335 Modelling Hierarchical Structure Between Dialogue Policy and Natural Language Generator with Option Framework for Task-oriented Dialogue System
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel algorithm called HDNO for policy optimization for task-oriented dialogue system so that the performance on the comprehensibility of generated responses is improved compared with other RL-based algorithms.		 Jianhong Wang; Yuan Zhang; Tae-Kyun Kim; Yunjie Gu;
336 Learning Explanations That Are Hard to Vary
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we investigate the principle that good explanations are hard to vary in the context of deep learning.		 Giambattista Parascandolo; Alexander Neitz; ANTONIO ORVIETO; Luigi Gresele; Bernhard Sch�lkopf;
337 Efficient Generalized Spherical CNNs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a generalized spherical CNN framework that encompasses various existing approaches and allows them to be leveraged alongside each other.		 Oliver Cobb; Christopher G. R. Wallis; Augustine N. Mavor-Parker; Augustin Marignier; Matthew A. Price; Mayeul d’Avezac; Jason McEwen;
338 Collective Robustness Certificates
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We fuse multiple single-prediction certificates into a drastically stronger collective certificate leveraging the locality property of Graph Neural Networks.		 Jan Schuchardt; Aleksandar Bojchevski; Johannes Klicpera; Stephan G�nnemann;
339 Entropic Gradient Descent Algorithms and Wide Flat Minima
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Relation between local entropy, flat minima, entropic algorithms and good generalization.		 Fabrizio Pittorino; Carlo Lucibello; Christoph Feinauer; Gabriele Perugini; Carlo Baldassi; Elizaveta Demyanenko; Riccardo Zecchina;
340 Achieving Linear Speedup with Partial Worker Participation in Non-IID Federated Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: So far, it remains an open question whether or not the linear speedup for convergence is achievable under non-i.i.d. datasets with partial worker participation in FL. In this paper, we show that the answer is affirmative.		 Haibo Yang; Minghong Fang; Jia Liu;
341 Categorical Normalizing Flows Via Continuous Transformations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we investigate Categorical Normalizing Flows, that is normalizing flows for categorical data.		 Phillip Lippe; Efstratios Gavves;
342 Learning to Represent Action Values As A Hypergraph on The Action Vertices
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper introduces a class of models for learning action representations in RL by leveraging the combinatorial structure of multidimensional discrete action spaces.		 Arash Tavakoli; Mehdi Fatemi; Petar Kormushev;
343 Debiasing Concept-based Explanations with Causal Analysis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use a technique from instrumental variables literature and remove the impact of noise and latent confounding from concept-based explanations.		 Mohammad Taha Bahadori; David Heckerman;
344 Lifelong Learning of Compositional Structures
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We create a general-purpose framework for lifelong learning of compositional structures that splits the learning process into two stages: assimilation of new tasks with existing components, and accommodation of new knowledge into the components.		 Jorge A Mendez; ERIC EATON;
345 Rethinking Embedding Coupling in Pre-trained Language Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Decoupling output embedding shapes leads to more transferable Transformer layers and prevents over-specialization of a Transformer’s upper layers to the pre-training task.		 Hyung Won Chung; Thibault Fevry; Henry Tsai; Melvin Johnson; Sebastian Ruder;
346 Creative Sketch Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce two creative sketch datasets and DoodlerGAN — a part-based GAN model that generates creative sketches.		 Songwei Ge; Vedanuj Goswami; Larry Zitnick; Devi Parikh;
347 Concept Learners for Few-Shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: COMET learns generalizable representations along human-understandable concept dimensions.		 Kaidi Cao; Maria Brbic; Jure Leskovec;
348 Domain Generalization with MixStyle
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: MixStyle makes CNNs more domain-generalizable by mixing instance-level feature statistics of training samples across domains.		 Kaiyang Zhou; Yongxin Yang; Yu Qiao; Tao Xiang;
349 DeLighT: Deep and Light-weight Transformer
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Deep and light-weight transformer that matches or improves the performance of baseline Transformers with 2 to 3 times fewer parameters on standard machine translation and language modeling tasks		 Sachin Mehta; Marjan Ghazvininejad; Srinivasan Iyer; Luke Zettlemoyer; Hannaneh Hajishirzi;
350 Single-Timescale Actor-Critic Provably Finds Globally Optimal Policy
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the global convergence and global optimality of actor-critic, one of the most popular families of reinforcement learning algorithms.		 Zuyue Fu; Zhuoran Yang; Zhaoran Wang;
351 Mastering Atari with Discrete World Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce DreamerV2, a reinforcement learning agent that learns behaviors purely from predictions in the compact latent space of a powerful world model.		 Danijar Hafner; Timothy P Lillicrap; Mohammad Norouzi; Jimmy Ba;
352 Learning Neural Event Functions for Ordinary Differential Equations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We discuss how event handling in ODE solvers can be differentiated through, allowing us to extend Neural ODEs to cases of implicitly defined termination times and enabling learning of discrete events and discontinuous dynamics.		 Ricky T. Q. Chen; Brandon Amos; Maximilian Nickel;
353 Contemplating Real-World Object Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We address whether current deep learning models are able to solve object recognition in real world and how robust they are to synthetic and natural distribution shifts.		 ali borji;
354 Neural Spatio-Temporal Point Processes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We motivate the use of Continuous-time Normalizing Flows for building spatio-temporal point processes, and discuss modeling conditional dependencies with recurrent- or attention-based Neural ODEs.		 Ricky T. Q. Chen; Brandon Amos; Maximilian Nickel;
355 Generative Time-series Modeling with Fourier Flows
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose an explicit likelihood model based on a novel class of normalizing flows that view time-series data in the frequency-domain rather than the time-domain.		 Ahmed Alaa; Alex James Chan; Mihaela van der Schaar;
356 DOP: Off-Policy Multi-Agent Decomposed Policy Gradients
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an off-policy multi-agent decomposed policy gradient method, addressing the drawbacks that prevent existing multi-agent policy gradient methods from achieving state-of-the-art performance.		 Yihan Wang; Beining Han; Tonghan Wang; Heng Dong; Chongjie Zhang;
357 The Risks of Invariant Risk Minimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We formally demonstrate that Invariant Risk Minimization and related alternative objectives often perform no better than standard ERM.		 Elan Rosenfeld; Pradeep Kumar Ravikumar; Andrej Risteski;
358 DynaTune: Dynamic Tensor Program Optimization in Deep Neural Network Compilation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We accelerate tensor program optimization by considering it as a multi-armed bandits problem and using Bayesian inference to achieve fast convergence.		 Minjia Zhang; Menghao Li; Chi Wang; Mingqin Li;
359 Bag of Tricks for Adversarial Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Empirical evaluation of basic training tricks used in adversarial training		 Tianyu Pang; Xiao Yang; Yinpeng Dong; Hang Su; Jun Zhu;
360 Learning with Instance-Dependent Label Noise: A Sample Sieve Approach
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper proposes a dynamic sample sieve method with strong theoretical guarantees to avoid overfitting to instance-based label noise.		 Hao Cheng; Zhaowei Zhu; Xingyu Li; Yifei Gong; Xing Sun; Yang Liu;
361 Efficient Reinforcement Learning in Factored MDPs with Application to Constrained RL
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an efficient algorithm with near-optimal regret guarantee for factored MDP, and apply the algorithm to a new formulation of constrained RL.		 Xiaoyu Chen; Jiachen Hu; Lihong Li; Liwei Wang;
362 Unbiased Teacher for Semi-Supervised Object Detection
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose Unbiased Teacher to jointly address the pseudo-labeling bias issue and the overfitting issue in semi-supervised object detection, and our model performs favorably against existing works on COCO-standard, COCO-additional, and VOC.		 Yen-Cheng Liu; Chih-Yao Ma; Zijian He; Chia-Wen Kuo; Kan Chen; Peizhao Zhang; Bichen Wu; Zsolt Kira; Peter Vajda;
363 Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A simple but effective nerual attention distillation method for backdoor defense.		 Yige Li; Xingjun Ma; Nodens Koren; Lingjuan Lyu; Xixiang Lyu; Bo Li;
364 Contrastive Learning with Adversarial Perturbations for Conditional Text Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a contrastive learning with adversarial perturbation to tackle the exposure bias problem.		 Seanie Lee; Dong Bok Lee; Sung Ju Hwang;
365 When Optimizing F-Divergence Is Robust with Label Noise
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show when maximizing a properly defined f-divergence measure with respect to a classifier’s predictions and the supervised labels is robust with label noise.		 Jiaheng Wei; Yang Liu;
366 Conditional Generative Modeling Via Learning The Latent Space
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Conditional generation in continuous multimodal spaces by learning the behavior of latent variables.		 Sameera Ramasinghe; Kanchana Nisal Ranasinghe; Salman Khan; Nick Barnes; Stephen Gould;
367 Text Generation By Learning from Off-Policy Demonstrations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose GOLD (generation by off-policy learning from demonstrations): an easy-to-optimize algorithm that learns from the off-policy demonstrations by importance weighting.		 Richard Yuanzhe Pang; He He;
368 Learning Long-term Visual Dynamics with Region Proposal Interaction Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose Region Proposal Interaction Networks for physical interaction prediction, which is applied across both simulation and real world environments for long-range prediction and planning.		 Haozhi Qi; Xiaolong Wang; Deepak Pathak; Yi Ma; Jitendra Malik;
369 ChipNet: Budget-Aware Pruning with Heaviside Continuous Approximations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A budget-aware deterministic strategy for structured pruning based on continuous Heaviside approximations and crispness loss.		 Rishabh Tiwari; Udbhav Bamba; Arnav Chavan; Deepak Gupta;
370 Learning to Deceive Knowledge Graph Augmented Models Via Targeted Perturbation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our findings raise doubts about KG-augmented models’ ability to leverage KG information and provide plausible explanations.		 Mrigank Raman; Hansen Wang; PeiFeng Wang; Siddhant Agarwal; Sungchul Kim; Ryan Rossi; Handong Zhao; Nedim Lipka; Xiang Ren;
371 IEPT: Instance-Level and Episode-Level Pretext Tasks for Few-Shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper proposes a novel Instance-level and Episode-level Pretext Task (IEPT) framework that seamlessly integrates SSL into FSL.		 Manli Zhang; Jianhong Zhang; Zhiwu Lu; Tao Xiang; Mingyu Ding; Songfang Huang;
372 The Role of Momentum Parameters in The Optimal Convergence of Adaptive Polyak’s Heavy-ball Methods
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A theory-practice gap in convex optimization and deep learning is bridged by giving a novel convergence analysis of the last Iterate of adaptive Heavy-ball methods.		 Wei Tao; Sheng Long; Gaowei Wu; Qing Tao;
373 Training with Quantization Noise for Extreme Model Compression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we extend this approach to work with extreme compression methods where the approximations introduced by STE are severe.		 Angela Fan; Pierre Stock; Benjamin Graham; Edouard Grave; R�mi Gribonval; Herve Jegou; Armand Joulin;
374 Adaptive Extra-Gradient Methods for Min-Max Optimization and Games
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop an adaptive mirror-prox method for min-max problems and games that achieves order-optimal rates in both smooth and non-smooth problems.		 Kimon Antonakopoulos; Veronica Belmega; Panayotis Mertikopoulos;
375 Distilling Knowledge from Reader to Retriever for Question Answering
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a technique to learn retriever models for downstream tasks, inspired by knowledge distillation, and which does not require annotated pairs of query and documents.		 Gautier Izacard; Edouard Grave;
376 Discovering Diverse Multi-Agent Strategic Behavior Via Reward Randomization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an MARL algorithm, RPG, which discovers diverse non-trivial strategic behavior in several challenging multi-agent games.		 Zhenggang Tang; Chao Yu; Boyuan Chen; Huazhe Xu; Xiaolong Wang; Fei Fang; Simon Shaolei Du; Yu Wang; Yi Wu;
377 Not-MIWAE: Deep Generative Modelling with Missing Not at Random Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an approach for building and fitting deep latent variable models (DLVMs) in cases where the missing process is dependent on the missing data.		 Niels Bruun Ipsen; Pierre-Alexandre Mattei; Jes Frellsen;
378 IDF++: Analyzing and Improving Integer Discrete Flows for Lossless Compression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We analyze and improve integer discrete normalizing flows for lossless source compression.		 Rianne van den Berg; Alexey A. Gritsenko; Mostafa Dehghani; Casper Kaae S�nderby; Tim Salimans;
379 Explaining By Imitating: Understanding Decisions By Interpretable Policy Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a method for learning interpretable representations of behavior to enable auditing, quantifying, and understanding human decision-making processes.		 Alihan H�y�k; Daniel Jarrett; Cem Tekin; Mihaela van der Schaar;
380 Learning with AMIGo: Adversarially Motivated Intrinsic Goals
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A constructively adversarial teacher-student setup can augment on-policy algorithms to better solve difficult exploration tasks in RL.		 Andres Campero; Roberta Raileanu; Heinrich Kuttler; Joshua B. Tenenbaum; Tim Rockt�schel; Edward Grefenstette;
381 Incorporating Symmetry Into Deep Dynamics Models for Improved Generalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Integrate various symmetris into deep sequence models for forecasting turbulence and ocean currents with improved accuracy and physical consistency.		 Rui Wang; Robin Walters; Rose Yu;
382 CaPC Learning: Confidential and Private Collaborative Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A method that enables parties to improve their own local heterogeneous machine learning models in a collaborative setting where both confidentiality and privacy need to be preserved to prevent both explicit and implicit sharing of private data.		 Christopher A. Choquette-Choo; Natalie Dullerud; Adam Dziedzic; Yunxiang Zhang; Somesh Jha; Nicolas Papernot; Xiao Wang;
383 Heating Up Decision Boundaries: Isocapacitory Saturation, Adversarial Scenarios and Generalization Bounds
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a heat-theoretic/Brownian motion approach to evaluate decision boundary geometry (curvature and density) with further applications in adversarial defenses, compression and generalization estimates.		 Bogdan Georgiev; Lukas Franken; Mayukh Mukherjee;
384 A PAC-Bayesian Approach to Generalization Bounds for Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we derived generalization bounds for two primary classes of graph neural networks (GNNs), namely graph convolutional networks (GCNs) and message passing GNNs, via a PAC-Bayesian approach.		 Renjie Liao; Raquel Urtasun; Richard Zemel;
385 Clairvoyance: A Pipeline Toolkit for Medical Time Series
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop and present Clairvoyance: a pipeline toolkit for medical time series.		 Daniel Jarrett; Jinsung Yoon; Ioana Bica; Zhaozhi Qian; Ari Ercole; Mihaela van der Schaar;
386 Self-supervised Representation Learning with Relative Predictive Coding
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present RPC, the Relative Predictive Coding, that achieves a good balance among the three challenges when modeling a contrastive learning objective: training stability, sensitivity to minibatch size, and downstream task performance.		 Yao-Hung Hubert Tsai; Martin Q. Ma; Muqiao Yang; Han Zhao; Louis-Philippe Morency; Ruslan Salakhutdinov;
387 Offline Model-Based Optimization Via Normalized Maximum Likelihood Estimation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Offline, data-driven optimization using normalized maximum likelihood to produce robust function estimates.		 Justin Fu; Sergey Levine;
388 On The Impossibility of Global Convergence in Multi-Loss Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove that desirable convergence properties cannot simultaneously hold for any multi-loss optimization algorithm.		 Alistair Letcher;
389 A Block Minifloat Representation for Training Deep Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new number representation, comparable to recently proposed 8-bit formats, for efficiently training a subset of DNN models.		 Sean Fox; Seyedramin Rasoulinezhad; Julian Faraone; david boland; Philip Leong;
390 Selectivity Considered Harmful: Evaluating The Causal Impact of Class Selectivity in DNNs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Class selectivity in CNNs is neither sufficient nor strictly necessary for optimal test accuracy		 Matthew L Leavitt; Ari S. Morcos;
391 Discrete Graph Structure Learning for Forecasting Multiple Time Series
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a graph neural network approach that learns a graph structure to enhance the forecasting of multiple multivariate time series.		 Chao Shang; Jie Chen; Jinbo Bi;
392 Contrastive Learning with Hard Negative Samples
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce an unsupervised method for sampling hard negatives for contrastive learning: the resulting embeddings have desirable theoretical properties, and have improved downstream performance on multiple different data modalities.		 Joshua David Robinson; Ching-Yao Chuang; Suvrit Sra; Stefanie Jegelka;
393 Intraclass Clustering: An Implicit Learning Ability That Regularizes DNNs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper provides empirical evidence that deep neural networks are implicitly regularized through their ability to extract meaningful clusters among the samples of a class.		 Simon Carbonnelle; Christophe De Vleeschouwer;
394 Sliced Kernelized Stein Discrepancy
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We proposed a method to tackle the curse-of-dimensionality issue of kernelized Stein discrepancy with RBF kernel, along with a novel particle inference algorithm resolving the vanishing repulsive issue of Stein variational gradient descent.		 Wenbo Gong; Yingzhen Li; Jos� Miguel Hern�ndez-Lobato;
395 Denoising Diffusion Implicit Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show and justify a GAN-like iterative generative model with relatively fast sampling, high sample quality and without any adversarial training.		 Jiaming Song; Chenlin Meng; Stefano Ermon;
396 Hierarchical Reinforcement Learning By Discovering Intrinsic Options
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Hierarchical RL that discovers short-horizon task-agnostic options to perform well on sparse reward manipulation and navigation tasks.		 Jesse Zhang; Haonan Yu; Wei Xu;
397 Answering Complex Open-Domain Questions with Multi-Hop Dense Retrieval
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a simple yet effective multi-hop dense retrieval approach for answering complex open-domain questions.		 Wenhan Xiong; Xiang Li; Srini Iyer; Jingfei Du; Patrick Lewis; William Yang Wang; Yashar Mehdad; Scott Yih; Sebastian Riedel; Douwe Kiela; Barlas Oguz;
398 Rethinking Soft Labels for Knowledge Distillation: A Bias�Variance Tradeoff Perspective
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: For knowledge distillation, we analyze the regularization effect introduced by soft labels from a bias-variance perspective and propose weighted soft labels to handle the tradeoff.		 Helong Zhou; Liangchen Song; Jiajie Chen; Ye Zhou; Guoli Wang; Junsong Yuan; Qian Zhang;
399 A Design Space Study for LISTA and Beyond
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Among all possible variations, we focus on extensively varying the connectivity patterns and neuron types, leading to a gigantic design space arising from LISTA.		 Tianjian Meng; Xiaohan Chen; Yifan Jiang; Zhangyang Wang;
400 What Should Not Be Contrastive in Contrastive Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a contrastive learning framework which does not require prior knowledge of specific, task-dependent invariances.		 Tete Xiao; Xiaolong Wang; Alexei A Efros; Trevor Darrell;
401 Do Wide and Deep Networks Learn The Same Things? Uncovering How Neural Network Representations Vary with Width and Depth
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that depth/width variations result in distinctive characteristics in the model internal representations, with resulting consequences for representations and output predictions across different model initializations and architectures.		 Thao Nguyen; Maithra Raghu; Simon Kornblith;
402 Learning to Set Waypoints for Audio-Visual Navigation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a hierarchical reinforcement learning approach to audio-visual navigation that learns to dynamically set waypoints in an end-to-end fashion		 Changan Chen; Sagnik Majumder; Ziad Al-Halah; Ruohan Gao; Santhosh Kumar Ramakrishnan; Kristen Grauman;
403 Semi-supervised Keypoint Localization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel method for semi-supervised keypoint localization via learning semantic keypoint representations.		 Olga Moskvyak; Frederic Maire; Feras Dayoub; Mahsa Baktashmotlagh;
404 Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our TE-NAS framework analyzes the spectrum of the neural tangent kernel (NTK) and the number of linear regions in the input space, achieving high-quality architecture search while dramatically reducing the search cost to four hours on ImageNet.		 Wuyang Chen; Xinyu Gong; Zhangyang Wang;
405 Off-Dynamics Reinforcement Learning: Training for Transfer with Domain Classifiers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a method for addressing domain adaptation in RL by using a (learned) modified reward, and prove that our method recovers a near-optimal policy for the target domain.		 Benjamin Eysenbach; Shreyas Chaudhari; Swapnil Asawa; Sergey Levine; Ruslan Salakhutdinov;
406 Federated Semi-Supervised Learning with Inter-Client Consistency & Disjoint Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a new practical problem of federated learning with a deficiency of supervision and study two realistic scenarios with a novel method to tackle the problems, including inter-client consistency and disjoint learning.		 Wonyong Jeong; Jaehong Yoon; Eunho Yang; Sung Ju Hwang;
407 Learning Safe Multi-agent Control with Decentralized Neural Barrier Certificates
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a safe and remarkably scalable multi-agent control approach via jointly learning the policy and decentralized control barrier certificates.		 Zengyi Qin; Kaiqing Zhang; Yuxiao Chen; Jingkai Chen; Chuchu Fan;
408 Direction Matters: On The Implicit Regularization Effect of Stochastic Gradient Descent with Moderate Learning Rate
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show a directional regularization effect for SGD with moderate learning rate		 Jingfeng Wu; Difan Zou; Vladimir Braverman; Quanquan Gu;
409 Meta Attention Networks: Meta-Learning Attention to Modulate Information Between Recurrent Independent Mechanisms
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Different time scale learning of independent mechanisms can lead to a better generalization.		 Kanika Madan; Nan Rosemary Ke; Anirudh Goyal; Bernhard Sch�lkopf; Yoshua Bengio;
410 Policy-Driven Attack: Learning to Query for Hard-label Black-box Adversarial Examples
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel hard-label black-box adversarial attack that introduces a reinforcement learning based formulation with a pre-trained policy network		 Ziang Yan; Yiwen Guo; Jian Liang; Changshui Zhang;
411 A Mathematical Exploration of Why Language Models Help Solve Downstream Tasks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a mathematical framework for understanding why language model features help with downstream linear classification tasks of interest		 Nikunj Saunshi; Sadhika Malladi; Sanjeev Arora;
412 Representation Learning for Sequence Data with Deep Autoencoding Predictive Components
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose Deep Autoencoding Predictive Components (DAPC) — a self-supervised representation learning method for sequence data, based on the intuition that useful representations of sequence data should exhibit a simple structure in the latent space.		 Junwen Bai; Weiran Wang; Yingbo Zhou; Caiming Xiong;
413 A Unifying View on Implicit Bias in Training Linear Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a unifying framework for analyzing implicit bias of linear networks and show theorems that extend existing results with less convergence assumptions.		 Chulhee Yun; Shankar Krishnan; Hossein Mobahi;
414 What Makes Instance Discrimination Good for Transfer Learning?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Understanding why self-supervised contrastive learning outperforms supervised counterparts for image pretraining		 Nanxuan Zhao; Zhirong Wu; Rynson W. H. Lau; Stephen Lin;
415 Learning Accurate Entropy Model with Global Reference for Image Compression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a novel Reference-based Model for image compression to effectively leverage both the local and global context information, which yields an enhanced compression performance.		 Yichen Qian; Zhiyu Tan; Xiuyu Sun; Ming Lin; Dongyang Li; Zhenhong Sun; Li Hao; Rong Jin;
416 Loss Function Discovery for Object Detection Via Convergence-Simulation Driven Search
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an effective convergence-simulation driven evolutionary search algorithm, called CSE-Autoloss, for object detection loss function discovery, which achieves 20x speedup via progressive convergence-simulation modules.		 Peidong Liu; Gengwei Zhang; Bochao Wang; Hang Xu; Xiaodan Liang; Yong Jiang; Zhenguo Li;
417 Effective Abstract Reasoning with Dual-Contrast Network
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a simple yet effective Dual-Contrast Network (DCNet) to solve Raven’s progressive matrices without using auxiliary annotations and assumptions.		 Tao Zhuo; Mohan Kankanhalli;
418 Do Not Let Privacy Overbill Utility: Gradient Embedding Perturbation for Private Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new algorithm for differentially private learning that advances state-of-the-art performance on several benchmark datasets.		 Da Yu; Huishuai Zhang; Wei Chen; Tie-Yan Liu;
419 Set Prediction Without Imposing Structure As Conditional Density Estimation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A set prediction training & prediction framework that addresses tasks with ambiguous predictions.		 David W Zhang; Gertjan J. Burghouts; Cees G. M. Snoek;
420 Clustering-friendly Representation Learning Via Instance Discrimination and Feature Decorrelation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a clustering-friendly representation learning method using instance discrimination and feature decorrelation, which achieves accuracy of 81.5% and 95.4% on CIFAR-10 and ImageNet-10, respectively, far above state-of-the-art values.		 Yaling Tao; Kentaro Takagi; Kouta Nakata;
421 Language-Agnostic Representation Learning of Source Code from Structure and Context
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Language-agnostic learning from Structure and Context of programs improves learning.		 Daniel Z�gner; Tobias Kirschstein; Michele Catasta; Jure Leskovec; Stephan G�nnemann;
422 Training GANs with Stronger Augmentations Via Contrastive Discriminator
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel discriminator of GAN showing that contrastive representation learning, e.g., SimCLR, and GAN can benefit each other when they are jointly trained.		 Jongheon Jeong; Jinwoo Shin;
423 Influence Functions in Deep Learning Are Fragile
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: End-to-end investigation of the behaviour of influence functions in deep learning		 Samyadeep Basu; Phil Pope; Soheil Feizi;
424 Separation and Concentration in Deep Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Numerical experiments demonstrate that deep neural network classifiers progressively separate class distributions around their mean, achieving linear separability on the training set, and increasing the Fisher discriminant ratio. We explain this mechanism with two types of operators.		 John Zarka; Florentin Guth; St�phane Mallat;
425 Colorization Transformer
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Self-attention for colorization		 Manoj Kumar; Dirk Weissenborn; Nal Kalchbrenner;
426 Autoregressive Dynamics Models for Offline Policy Evaluation and Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We demonstrate autoregressive dynamics models outperform standard feedforward models and other baselines in offline policy evaluation and optimization.		 Michael R Zhang; Thomas Paine; Ofir Nachum; Cosmin Paduraru; George Tucker; ziyu wang; Mohammad Norouzi;
427 FedBN: Federated Learning on Non-IID Features Via Local Batch Normalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel and efficient federated learning aggregation method, denoted FedBN, that uses local batch normalization to effectively tackle the underexplored non-iid problem of heterogeneous feature distributions, or feature shift.		 Xiaoxiao Li; Meirui JIANG; Xiaofei Zhang; Michael Kamp; Qi Dou;
428 Learning Robust State Abstractions for Hidden-Parameter Block MDPs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we leverage ideas of common structure from the HiP-MDP setting, and extend it to enable robust state abstractions inspired by Block MDPs.		 Amy Zhang; Shagun Sodhani; Khimya Khetarpal; Joelle Pineau;
429 Meta-Learning with Neural Tangent Kernels
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: First work to define meta learning in RKHS induced by Neural Tangent Kernel		 Yufan Zhou; Zhenyi Wang; Jiayi Xian; Changyou Chen; Jinhui Xu;
430 Continual Learning in Recurrent Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper studies the behavior of established approaches to the problem of continual learning in the context of recurrent neural networks.		 Benjamin Ehret; Christian Henning; Maria Cervera; Alexander Meulemans; Johannes Von Oswald; Benjamin F Grewe;
431 A Trainable Optimal Transport Embedding for Feature Aggregation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new, trainable embedding for large sets of features such as biological sequences, and demonstrate its effectiveness.		 Gr�goire Mialon; Dexiong Chen; Alexandre d’Aspremont; Julien Mairal;
432 Learning What-if Explanations for Sequential Decision-Making
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose explaining sequential decision-making by integrating counterfactual reasoning into batch inverse reinforcement learning and recovering the preferences of experts over what-if outcomes.		 Ioana Bica; Daniel Jarrett; Alihan H�y�k; Mihaela van der Schaar;
433 Improving Transformation Invariance in Contrastive Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose methods to strengthen the invariance properties of representations obtained by contrastive learning using novel gradient regularization during training and feature averaging at test time. Finally, we introduce the novel Spirograph dataset to explore our ideas in the context of a differentiable generative process with multiple downstream tasks, showing that our techniques for learning invariance are highly beneficial.		 Adam Foster; Rattana Pukdee; Tom Rainforth;
434 Shapley Explainability on The Data Manifold
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present drawbacks of model explanations that do not respect the data manifold, and introduce two methods for on-manifold explainability.		 Christopher Frye; Damien de Mijolla; Tom Begley; Laurence Cowton; Megan Stanley; Ilya Feige;
435 Noise Or Signal: The Role of Image Backgrounds in Object Recognition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop and use a toolkit to investigate models? use of (and reliance on) image backgrounds.		 Kai Yuanqing Xiao; Logan Engstrom; Andrew Ilyas; Aleksander Madry;
436 Enjoy Your Editing: Controllable GANs for Image Editing Via Latent Space Navigation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a state-of-the-art approach to semantically edit images by transferring latent vectors towards meaningful latent space directions.		 Peiye Zhuang; Oluwasanmi O Koyejo; Alex Schwing;
437 Perceptual Adversarial Robustness: Generalizable Defenses Against Unforeseen Threat Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Adversarial training against a perceptually-aligned attack gives high robustness against many diverse adversarial threat models.		 Cassidy Laidlaw; Sahil Singla; Soheil Feizi;
438 Zero-Cost Proxies for Lightweight NAS
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A single minibatch of data is used to score neural networks for NAS instead of performing full training.		 Mohamed S Abdelfattah; Abhinav Mehrotra; Lukasz Dudziak; Nicholas Donald Lane;
439 Usable Information and Evolution of Optimal Representations During Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a notion of usable information contained in the representation learned by a deep network, and use it to study how optimal representations for the task emerge during training, and how they adapt to different tasks.		 Michael Kleinman; Alessandro Achille; Daksh Idnani; Jonathan Kao;
440 Exploring The Uncertainty Properties of Neural Networks� Implicit Priors in The Infinite-Width Limit
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the uncertainty properties of infinitely-wide neural networks		 Ben Adlam; Jaehoon Lee; Lechao Xiao; Jeffrey Pennington; Jasper Snoek;
441 On The Geometry of Generalization and Memorization in Deep Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We analyze the representational geometry of deep neural networks during generalization and memorization, and find the structure across layers of when and where memorization occurs, as well as drivers for this emerging structure.		 Cory Stephenson; suchismita padhy; Abhinav Ganesh; Yue Hui; Hanlin Tang; SueYeon Chung;
442 Deep Partition Aggregation: Provable Defenses Against General Poisoning Attacks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose novel certified defenses against label-flipping and general adversarial poisoning attacks.		 Alexander Levine; Soheil Feizi;
443 DC3: A Learning Method for Optimization with Hard Constraints
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We describe a method, DC3, for fast approximate solutions to optimization problems with hard constraints, which enforces feasibility via a differentiable procedure incorporated into a neural network.		 Priya L. Donti; David Rolnick; J Zico Kolter;
444 Is Label Smoothing Truly Incompatible with Knowledge Distillation: An Empirical Study
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work empirically clarifies a recently discovered perspective that label smoothing is incompatible with knowledge distillation.		 Zhiqiang Shen; Zechun Liu; Dejia Xu; Zitian Chen; Kwang-Ting Cheng; Marios Savvides;
445 Shape-Texture Debiased Neural Network Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Training CNNs to acquire a debiased shape-texture representation improves image recognition.		 Yingwei Li; Qihang Yu; Mingxing Tan; Jieru Mei; Peng Tang; Wei Shen; Alan Yuille; cihang xie;
446 Using Latent Space Regression to Analyze and Leverage Compositionality in GANs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use a latent regressor network to investigate compositional properties of image synthesis with GANs.		 Lucy Chai; Jonas Wulff; Phillip Isola;
447 Blending MPC & Value Function Approximation for Efficient Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A framework for blending model-predictive control and model-free value function learning to systematically trade-off bias due to approximate dynamics models and value functions learned from real data		 Mohak Bhardwaj; Sanjiban Choudhury; Byron Boots;
448 Model Patching: Closing The Subgroup Performance Gap with Data Augmentation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We describe how to fix classifiers that fail on subgroups of a class using a combination of learned data augmentation & consistency training to achieve subgroup invariance.		 Karan Goel; Albert Gu; Yixuan Li; Christopher Re;
449 Non-asymptotic Confidence Intervals of Off-policy Evaluation: Primal and Dual Bounds
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an approach to constructing non-asymptotic confidence intervals of off-policy estimation.		 Yihao Feng; Ziyang Tang; na zhang; qiang liu;
450 Linear Mode Connectivity in Multitask and Continual Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that continual and multitask minima are connected by linear low-error paths and design an effective continual learning algorithm that exploits this property.		 Seyed Iman Mirzadeh; Mehrdad Farajtabar; Dilan Gorur; Razvan Pascanu; Hassan Ghasemzadeh;
451 Robust and Generalizable Visual Representation Learning Via Random Convolutions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use random convolutions as data augmentation to train robust visual representation that generalize to new domains.		 Zhenlin Xu; Deyi Liu; Junlin Yang; Colin Raffel; Marc Niethammer;
452 Intrinsic-Extrinsic Convolution and Pooling for Learning on 3D Protein Structures
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a new neural network architecture to process 3D protein structures. Second, we introduce a set of hierarchical pooling operators that enable multi-scale protein analysis.		 Pedro Hermosilla; Marco Sch�fer; Matej Lang; Gloria Fackelmann; Pere-Pau V�zquez; Barbora Kozlikova; Michael Krone; Tobias Ritschel; Timo Ropinski;
453 Variational State-Space Models for Localisation and Dense 3D Mapping in 6 DoF
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a variational state-space model with a latent map for 6-DoF localisation, 3D dense mapping and generative modelling for planning.		 Atanas Mirchev; Baris Kayalibay; Patrick van der Smagt; Justin Bayer;
454 AdamP: Slowing Down The Slowdown for Momentum Optimizers on Scale-invariant Weights
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we verify that the widely-adopted combination of the two ingredients lead to the premature decay of effective step sizes and sub-optimal model performances.		 Byeongho Heo; Sanghyuk Chun; Seong Joon Oh; Dongyoon Han; Sangdoo Yun; Gyuwan Kim; Youngjung Uh; Jung-Woo Ha;
455 MiCE: Mixture of Contrastive Experts for Unsupervised Image Clustering
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A principled probabilistic clustering method that exploits the discriminative representations learned by contrastive learning and the semantic structures captured by a latent mixture model in a unified framework.		 Tsung Wei Tsai; Chongxuan Li; Jun Zhu;
456 HalentNet: Multimodal Trajectory Forecasting with Hallucinative Intents
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Towards this goal, we introduce HalentNet to better model the future motion distribution in addition to a traditional trajectory regression learning objective by incorporating generative augmentation losses.		 Deyao Zhu; Mohamed Zahran; Li Erran Li; Mohamed Elhoseiny;
457 Model-based Micro-data Reinforcement Learning: What Are The Crucial Model Properties and Which Model to Choose?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Crucial model properties for model-based reinforcement learning: multi-modal posterior predictives and heteroscedasticity.		 Bal�zs K�gl; Gabriel Hurtado; Albert Thomas;
458 Private Image Reconstruction from System Side Channels Using Generative Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present the first generative model-based side channel analysis (SCA) to reconstruct private user images.		 Yuanyuan Yuan; Shuai Wang; Junping Zhang;
459 Contextual Transformation Networks for Online Continual Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper develops a novel method that can model task-specific features with minimal complexity overhead.		 Quang Pham; Chenghao Liu; Doyen Sahoo; Steven HOI;
460 Towards A Unified Understanding and Improving of Adversarial Transferability
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove the close relationship between the interaction and adversarial transferability, provide a unified explanation for previous transferability-boosting methods, and develop a loss to improve adversarial transferability.		 Xin Wang; Jie Ren; Shuyun Lin; Xiangming Zhu; Yisen Wang; Quanshi Zhang;
461 The Inductive Bias of ReLU Networks on Orthogonally Separable Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We characterise the function learnt by two-layer ReLU nets trained on orthogonally separable data.		 Mary Phuong; Christoph H Lampert;
462 A Statistical Theory of Cold Posteriors in Deep Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a generative model of dataset curation that explains the cold-posterior effect		 Laurence Aitchison;
463 IOT: Instance-wise Layer Reordering for Transformer Structures
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Based on this observation, in this work, we break the assumption of the fixed layer order in Transformer and introduce instance-wise layer reordering into model structure.		 Jinhua Zhu; Lijun Wu; Yingce Xia; Shufang Xie; Tao Qin; Wengang Zhou; Houqiang Li; Tie-Yan Liu;
464 Counterfactual Generative Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A generative model structured into independent causal mechanisms produces images for training invariant classifiers.		 Axel Sauer; Andreas Geiger;
465 Conditionally Adaptive Multi-Task Learning: Improving Transfer Learning in NLP Using Fewer Parameters & Less Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Can multi-task outperform single task fine-tuning? CA-MTL is a new method that shows that it is possible with task conditioned model adaption and uncertainty sampling.		 Jonathan Pilault; Amine El hattami; Christopher Pal;
466 Towards Impartial Multi-task Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an impartial multi-task learning method that treats all tasks equally without bias towards any task.		 Liyang Liu; Yi Li; Zhanghui Kuang; Jing-Hao Xue; Yimin Chen; Wenming Yang; Qingmin Liao; Wayne Zhang;
467 Theoretical Bounds on Estimation Error for Meta-learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove novel minimax risk lower bounds and upper bounds for meta learners		 James Lucas; Mengye Ren; Irene Raissa KAMENI KAMENI; Toniann Pitassi; Richard Zemel;
468 Domain-Robust Visual Imitation Learning with Mutual Information Constraints
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Imitation of visual expert demonstrations robust to appearance and embodiment mismatch, working for high dimensional control problems.		 Edoardo Cetin; Oya Celiktutan;
469 Unsupervised Representation Learning for Time Series with Temporal Neighborhood Coding
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An unsupervised representation learning framework for high-dimensional non-stationary time series		 Sana Tonekaboni; Danny Eytan; Anna Goldenberg;
470 Enforcing Robust Control Guarantees Within Neural Network Policies
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a generic nonlinear control policy class, parameterized by neural networks, that nonetheless enforces the same provable robustness criteria as robust control.		 Priya L. Donti; Melrose Roderick; Mahyar Fazlyab; J Zico Kolter;
471 Active Contrastive Learning of Audio-Visual Video Representations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an active learning approach to improve negative sampling for contrastive learning and demonstrate it on learning audio-visual representations from videos.		 Shuang Ma; Zhaoyang Zeng; Daniel McDuff; Yale Song;
472 Parameter Efficient Multimodal Transformers for Video Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a technique to reduce the number of parameters in multimodal BERT models up to 80% (from 155 million to 31 million parameters).		 Sangho Lee; Youngjae Yu; Gunhee Kim; Thomas Breuel; Jan Kautz; Yale Song;
473 Robust Pruning at Initialization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Making pruning at initialization robust to Gradient vanishing/exploding		 Soufiane Hayou; Jean-Francois Ton; Arnaud Doucet; Yee Whye Teh;
474 Efficient Wasserstein Natural Gradients for Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop novel, efficient estimators for the Wasserstein natural gradient applied to reinforcement learning that improve the efficiency and performance of advanced baselines.		 Ted Moskovitz; Michael Arbel; Ferenc Huszar; Arthur Gretton;
475 Probing BERT in Hyperbolic Spaces
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a Poincar? probe for finding syntax and sentiments from BERT in hyperbolic spaces.		 Boli Chen; Yao Fu; Guangwei Xu; Pengjun Xie; Chuanqi Tan; Mosha Chen; Liping Jing;
476 On Fast Adversarial Robustness Adaptation in Model-Agnostic Meta-Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a general but easily-optimized robustness-regularized meta-learning framework, which allows the use of unlabeled data augmentation, fast adversarial attack generation, and computationally-light fine-tuning.		 Ren Wang; Kaidi Xu; Sijia Liu; Pin-Yu Chen; Tsui-Wei Weng; Chuang Gan; Meng Wang;
477 Anatomy of Catastrophic Forgetting: Hidden Representations and Task Semantics
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the layerwise change in representations due to catastrophic forgetting, and use our understanding to study how task similarity influences forgetting.		 Vinay Venkatesh Ramasesh; Ethan Dyer; Maithra Raghu;
478 Trusted Multi-View Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To this end, we propose a novel multi-view classification method, termed trusted multi-view classification, which provides a new paradigm for multi-view learning by dynamically integrating different views at an evidence level.		 Zongbo Han; Changqing Zhang; Huazhu Fu; Joey Tianyi Zhou;
479 I-Mix: A Strategy for Regularizing Contrastive Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose i-Mix, a simple yet effective strategy for regularizing contrastive representation learning in both vision and non-vision domains.		 Kibok Lee; Yian Zhu; Kihyuk Sohn; Chun-Liang Li; Jinwoo Shin; Honglak Lee;
480 Initialization and Regularization of Factorized Neural Layers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Principled initialization and regularization of factorized neural layers leads to strong performance in compression, knowledge distillation, and language modeling tasks.		 Mikhail Khodak; Neil A. Tenenholtz; Lester Mackey; Nicolo Fusi;
481 Learning to Generate 3D Shapes with Generative Cellular Automata
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a Markov chain based 3D generative model named Generative Cellular Automata (GCA), that progressively evolves to a shape of the learned distribution using the local update rules of cellular automata.		 Dongsu Zhang; Changwoon Choi; Jeonghwan Kim; Young Min Kim;
482 Self-Supervised Learning of Compressed Video Representations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a self-supervised approach to learning compressed video representations.		 Youngjae Yu; Sangho Lee; Gunhee Kim; Yale Song;
483 Factorizing Declarative and Procedural Knowledge in Structured, Dynamical Environments
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We explore separate factorization of procedural and declarative knowledge in modular recurrent neural networks.		 Anirudh Goyal; Alex Lamb; Phanideep Gampa; Philippe Beaudoin; Charles Blundell; Sergey Levine; Yoshua Bengio; Michael Curtis Mozer;
484 Cut Out The Annotator, Keep The Cutout: Better Segmentation with Weak Supervision
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we present a weak supervision approach for segmentation tasks, allowing users to train high-performing segmentation CNNs with very few hand-labeled training points.		 Sarah Hooper; Michael Wornow; Ying Hang Seah; Peter Kellman; Hui Xue; Frederic Sala; Curtis Langlotz; Christopher Re;
485 FastSpeech 2: Fast and High-Quality End-to-End Text to Speech
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a non-autoregressive TTS model named FastSpeech 2 to better solve the one-to-many mapping problem in TTS and surpass autoregressive models in voice quality.		 Yi Ren; Chenxu Hu; Xu Tan; Tao Qin; Sheng Zhao; Zhou Zhao; Tie-Yan Liu;
486 On Learning Universal Representations Across Languages
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we extend pre-trained language models to learn universal representations among multiple languages, and show the effectiveness on cross-lingual understanding and generation.		 Xiangpeng Wei; Yue Hu; Rongxiang Weng; Luxi Xing; Heng Yu; Weihua Luo;
487 Effective Distributed Learning with Random Features: Improved Bounds and Algorithms
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This papaer focuses on the studies of the statistical properties of distributed KRR together with random features		 Yong Liu; Jiankun Liu; Shuqiang Wang;
488 Multi-Class Uncertainty Calibration Via Mutual Information Maximization-based Binning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose I-Max binning, a novel method for multi-class calibration, improving over previous methods in terms of various ECE measures		 Kanil Patel; William H. Beluch; Bin Yang; Michael Pfeiffer; Dan Zhang;
489 Neural ODE Processes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a stochastic process based on Neural ODEs.		 Alexander Norcliffe; Cristian Bodnar; Ben Day; Jacob Moss; Pietro Li�;
490 Conformation-Guided Molecular Representation with Hamiltonian Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a molecular representation algorithm, which preserves molecular conformations with a neural physics engine and generates fingerprints with an MPNN.		 Ziyao Li; Shuwen Yang; Guojie Song; Lingsheng Cai;
491 An Unsupervised Deep Learning Approach for Real-World Image Denoising
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an unsupervised real-world image denoising approach that combines DNNs with classical MAP approaches.		 Dihan Zheng; Sia Huat Tan; Xiaowen Zhang; Zuoqiang Shi; Kaisheng Ma; Chenglong Bao;
492 Uncertainty in Gradient Boosting Via Ensembles
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Propose and analyze an ensemble-based framework for deriving uncertainty estimates in GBDT models.		 Andrey Malinin; Liudmila Prokhorenkova; Aleksei Ustimenko;
493 Lossless Compression of Structured Convolutional Models Via Lifting
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Speeding up weight-sharing dynamic neural computation graphs, such as GNNs, with lifted inference.		 Gustav Sourek; Filip Zelezny; Ondrej Kuzelka;
494 Neural Networks with Late-phase Weights
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Here, we show that the solutions found by SGD can be further improved by ensembling a subset of the weights in late stages of learning.		 Johannes Von Oswald; Seijin Kobayashi; Joao Sacramento; Alexander Meulemans; Christian Henning; Benjamin F Grewe;
495 Disambiguating Symbolic Expressions in Informal Documents
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the task of \emph{disambiguating} symbolic expressions in informal STEM documents in the form of \LaTeX files — that is, determining their precise semantics and abstract syntax tree — as a neural machine translation task.		 Dennis M�ller; Cezary Kaliszyk;
496 Learning Parametrised Graph Shift Operators
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a parametrised graph shift operator (PGSO) to encode graph structure, providing a unified view of the most common GSOs, and improve GNN performance by incorporating the PGSO into the model training in an end-to-end manner.		 George Dasoulas; Johannes F. Lutzeyer; Michalis Vazirgiannis;
497 Efficient Conformal Prediction Via Cascaded Inference with Expanded Admission
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work proposes two complementary techniques for improving the efficiency of conformal prediction in large-scale domains—while still retaining performance guarantees.		 Adam Fisch; Tal Schuster; Tommi S. Jaakkola; Regina Barzilay;
498 GANs Can Play Lottery Tickets Too
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Winning tickets exist in the deep generative adversarial networks, which substantially outperform previous state-of-the-art compressed GAN		 Xuxi Chen; Zhenyu Zhang; Yongduo Sui; Tianlong Chen;
499 ResNet After All: Neural ODEs and Their Numerical Solution
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We explain why some Neural ODE models do not permit a continuous-depth interpretation after training and how to fix it.		 Katharina Ott; Prateek Katiyar; Philipp Hennig; Michael Tiemann;
500 Semantic Re-tuning with Contrastive Tension
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A self-supervised method for learning STS related sentence embedding using pre-trained language models, setting a new SOTA for STS related embedding tasks.		 Fredrik Carlsson; Magnus Sahlgren; Evangelia Gogoulou; Amaru Cuba Gyllensten; Erik Ylip�� Hellqvist;
501 Property Controllable Variational Autoencoder Via Invertible Mutual Dependence
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel generative model for learning interpretable latent representation for generating data with desired properties.		 Xiaojie Guo; Yuanqi Du; Liang Zhao;
502 Latent Convergent Cross Mapping
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Latent CCM uses reconstruction between latent processes of dynamical systems to infer causality between short and sporadic time series.		 Edward De Brouwer; Adam Arany; Jaak Simm; Yves Moreau;
503 Adaptive Universal Generalized PageRank Graph Neural Network
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We combine generalized PageRank with GNNs to adapt universal node label patterns and the over-smoothing problem.		 Eli Chien; Jianhao Peng; Pan Li; Olgica Milenkovic;
504 Neural Learning of One-of-Many Solutions for Combinatorial Problems in Structured Output Spaces
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work identifies and proposes a solution for handling solution multiplicity while learning neural methods for combinatorial problems in structured output spaces.		 Yatin Nandwani; Deepanshu Jindal; Mausam .; Parag Singla;
505 My Body Is A Cage: The Role of Morphology in Graph-Based Incompatible Control
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Transformer-based approach to multitask incompatible continuous control inspired by a hypothesis that any benefits GNNs extract from the graph structure are outweighed by difficulties they create for message passing.		 Vitaly Kurin; Maximilian Igl; Tim Rockt�schel; Wendelin Boehmer; Shimon Whiteson;
506 FedBE: Making Bayesian Model Ensemble Applicable to Federated Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a novel aggregation algorithm named FedBE, which takes a Bayesian inference perspective by sampling higher-quality global models and combining them via Bayesian model Ensemble, leading to much robust aggregation.		 Hong-You Chen; Wei-Lun Chao;
507 MALI: A Memory Efficient and Reverse Accurate Integrator for Neural ODEs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A solver for ODE that guarantees accuracy in reverse-time trajectory at a constant memory cost		 Juntang Zhuang; Nicha C Dvornek; sekhar tatikonda; James s Duncan;
508 Reducing The Computational Cost of Deep Generative Models with Binary Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We demonstrate that deep generative models can be effectively trained using binary weights and/or activations		 Thomas Bird; Friso Kingma; David Barber;
509 In-N-Out: Pre-Training and Self-Training Using Auxiliary Information for Out-of-Distribution Robustness
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Using auxiliary information as inputs hurts OOD, but using auxiliary information by pretraining and self-training improves in-distribution and OOD accuracies on real-world datasets, with theoretical guarantees in a linear multi-task setting.		 Sang Michael Xie; Ananya Kumar; Robbie Jones; Fereshte Khani; Tengyu Ma; Percy Liang;
510 Incremental Few-shot Learning Via Vector Quantization in Deep Embedded Space
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this study, we propose a nonparametric method in deep embedded space to tackle incremental few-shot learning problems.		 Kuilin Chen; Chi-Guhn Lee;
511 Contrastive Syn-to-Real Generalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new contrastive synthetic-to-real generalization framework that achieves state-of-the-art performance on synthetic-to-real generalization problems.		 Wuyang Chen; Zhiding Yu; Shalini De Mello; Sifei Liu; Jose M. Alvarez; Zhangyang Wang; Anima Anandkumar;
512 Remembering for The Right Reasons: Explanations Reduce Catastrophic Forgetting
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Introducing a connection between continual learning and model explainability by regularizing saliency maps to avoid forgetting and showing its effect on memory and regularization-based continual learning approaches.		 Sayna Ebrahimi; Suzanne Petryk; Akash Gokul; William Gan; Joseph E. Gonzalez; Marcus Rohrbach; trevor darrell;
513 Leaky Tiling Activations: A Simple Approach to Learning Sparse Representations Online
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A simple and efficient way to learn sparse feature in deep learning setting.		 Yangchen Pan; Kirby Banman; Martha White;
514 High-Capacity Expert Binary Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: (a) To increase model capacity, we propose Expert Binary Convolution, which, for the first time, tailors conditional computing to binary networks by learning to select one data-specific expert binary filter at a time conditioned on input features.		 Adrian Bulat; Brais Martinez; Georgios Tzimiropoulos;
515 Learning What To Do By Simulating The Past
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Imitating policies given just a single state sampled from a rollout from an expert		 David Lindner; Rohin Shah; Pieter Abbeel; Anca Dragan;
516 Progressive Skeletonization: Trimming More Fat from A Network at Initialization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We find performance of current methods for pruning at initialization plummets at high sparsity levels, we study the possible reasons and present a more robust method overall.		 Pau de Jorge; Amartya Sanyal; Harkirat Behl; Philip Torr; Gr�gory Rogez; Puneet K. Dokania;
517 Filtered Inner Product Projection for Multilingual Embedding Alignment
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a method, Filtered Inner Product Projection (FIPP), for mapping embeddings to a common representation space.		 Vin Sachidananda; Ziyi Yang; Chenguang Zhu;
518 Learning Manifold Patch-Based Representations of Man-Made Shapes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a parametrically defined patch-based 3D shape representation that is compatible both with traditional CAD modeling tools and modern deep learning pipelines.		 Dmitriy Smirnov; Mikhail Bessmeltsev; Justin Solomon;
519 Aligning AI With Shared Human Values
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We approach a longstanding problem in machine ethics provide evidence that it is soluble.		 Dan Hendrycks; Collin Burns; Steven Basart; Andrew Critch; Jerry Li; Dawn Song; Jacob Steinhardt;
520 Kanerva++: Extending The Kanerva Machine With Differentiable, Locally Block Allocated Latent Memory
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Differentiable block allocated latent memory model for generative modeling.		 Jason Ramapuram; Yan Wu; Alexandros Kalousis;
521 Measuring Massive Multitask Language Understanding
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We test language models on 57 different multiple-choice tasks.		 Dan Hendrycks; Collin Burns; Steven Basart; Andy Zou; Mantas Mazeika; Dawn Song; Jacob Steinhardt;
522 Towards Robust Neural Networks Via Close-loop Control
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a close-loop control framework to improve the robustness of neural networks under various data perturbations.		 Zhuotong Chen; Qianxiao Li; Zheng Zhang;
523 Statistical Inference for Individual Fairness
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we focus on the problem of detecting violations of individual fairness in ML models.		 Subha Maity; Songkai Xue; Mikhail Yurochkin; Yuekai Sun;
524 HyperGrid Transformers: Towards A Single Model for Multiple Tasks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: State-of-the-art multi-task NLU performance with only a single model		 Yi Tay; Zhe Zhao; Dara Bahri; Donald Metzler; Da-Cheng Juan;
525 Greedy-GQ with Variance Reduction: Finite-time Analysis and Improved Complexity
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a variance-reduced Greedy-GQ (VR-Greedy-GQ) algorithm for off-policy optimal control.		 Shaocong Ma; Ziyi Chen; Yi Zhou; Shaofeng Zou;
526 On InstaHide, Phase Retrieval, and Sparse Matrix Factorization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We examine the security of InstaHide, a recently proposed framework for private distributed learning, through the lens of phase retrieval and give an attack when the underlying datasets are Gaussian.		 Sitan Chen; Xiaoxiao Li; Zhao Song; Danyang Zhuo;
527 VA-RED2: Video Adaptive Redundancy Reduction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Here we present a redundancy reduction framework, termed VA-RED2, which is {\em input-dependent}.		 Bowen Pan; Rameswar Panda; Camilo Luciano Fosco; Chung-Ching Lin; Alex J Andonian; Yue Meng; Kate Saenko; Aude Oliva; Rogerio Feris;
528 SEDONA: Search for Decoupled Neural Networks Toward Greedy Block-wise Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our approach is the first attempt to automate decoupling neural networks for greedy block-wise learning and outperforms both end-to-end backprop and state-of-the-art greedy-learning methods on CIFAR-10, Tiny-ImageNet and ImageNet classification.		 Myeongjang Pyeon; Jihwan Moon; Taeyoung Hahn; Gunhee Kim;
529 ALFWorld: Aligning Text and Embodied Environments for Interactive Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Pre-training in text-based environments allows agents to learn priors and policies that help solve embodied tasks.		 Mohit Shridhar; Xingdi Yuan; Marc-Alexandre Cote; Yonatan Bisk; Adam Trischler; Matthew Hausknecht;
530 Learning Task Decomposition with Order-Memory Policy Network
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce an Ordered Memory Policy Network (OMPN) to discover task decomposition by imitation learning from demonstration.		 Yuchen Lu; Yikang Shen; Siyuan Zhou; Aaron Courville; Joshua B. Tenenbaum; Chuang Gan;
531 Adversarially-Trained Deep Nets Transfer Better
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We demonstrate that adversarially-trained models transfer better to new domains than naturally-trained models, especially when only limited training data is available in the target domain.		 Francisco Utrera; Evan Kravitz; N. Benjamin Erichson; Rajiv Khanna; Michael W. Mahoney;
532 UMEC: Unified Model and Embedding Compression for Efficient Recommendation Systems
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a unified model and embedding compression (UMEC) framework to hammer an efficient neural network-based recommendation system.		 Jiayi Shen; Haotao Wang; Shupeng Gui; Jianchao Tan; Zhangyang Wang; Ji Liu;
533 Exploring Balanced Feature Spaces for Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Inspired by these insights, we develop a novel representation learning method, called k-positive contrastive learning.		 Bingyi Kang; Yu Li; Sa Xie; Zehuan Yuan; Jiashi Feng;
534 Calibration of Neural Networks Using Splines
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a binning-free calibration measure inspired by the classical Kolmogorov-Smirnov statistical test and obtain a recalibration function by approximating the empirical cumulative distribution using a differentiable function via splines.		 Kartik Gupta; Amir Rahimi; Thalaiyasingam Ajanthan; Thomas Mensink; Cristian Sminchisescu; Richard Hartley;
535 Improving Relational Regularized Autoencoders with Spherical Sliced Fused Gromov Wasserstein
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Improving relational regularized autoencoder by introducing new sliced optimal transport discrepancies between the prior and aggregated posterior distributions.		 Khai Nguyen; Son Nguyen; Nhat Ho; Tung Pham; Hung Bui;
536 Rethinking Positional Encoding in Language Pre-training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel and better positional encoding method for Transformer-based language pre-training models.		 Guolin Ke; Di He; Tie-Yan Liu;
537 Discovering Autoregressive Orderings with Variational Inference
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The paper proposes an unsupervised learner that discovers high-quality autoregressive orders for language generation without domain-specific prior.		 Xuanlin Li; Brandon Trabucco; Dong Huk Park; Yang Gao; Michael Luo; Sheng Shen; Trevor Darrell;
538 Differentiable Trust Region Layers for Deep Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose differentiable neural network layers to enforce trust regions for deep Gaussian policies via closed-form projections.		 Fabian Otto; Philipp Becker; Vien Anh Ngo; Hanna Carolin Maria Ziesche; Gerhard Neumann;
539 SaliencyMix: A Saliency Guided Data Augmentation Strategy for Better Regularization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The proposed method carefully selects a representative image patch with the help of a saliency map and mixes this indicative patch with the target image that leads the model to learn more appropriate feature representation		 A F M Shahab Uddin; Mst. Sirazam Monira; Wheemyung Shin; TaeChoong Chung; Sung-Ho Bae;
540 Task-Agnostic Morphology Evolution
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce TAME, a novel method for optimizing agent morphology using only randomly sampled action primitives and no task driven reward signals.		 Donald Joseph Hejna III; Pieter Abbeel; Lerrel Pinto;
541 Learning Associative Inference Using Fast Weight Memory
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a Recurrent Neural Network model which is augmented with an associative memory to generalise in a more systematically		 Imanol Schlag; Tsendsuren Munkhdalai; J�rgen Schmidhuber;
542 Boost Then Convolve: Gradient Boosting Meets Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel strong architecture that combines advantages of GBDT and GNN for node-level prediction problems on graphs with tabular data.		 Sergei Ivanov; Liudmila Prokhorenkova;
543 Degree-Quant: Quantization-Aware Training for Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We provide a training technique that enables graph neural networks to use low precision integer arithmetic at inference time, yielding up to 4.7x latency improvements on CPU		 Shyam Anil Tailor; Javier Fernandez-Marques; Nicholas Donald Lane;
544 Network Pruning That Matters: A Case Study on Retraining Variants
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the effective of different retraining mechanisms while doing pruning		 Duong Hoang Le; Binh-Son Hua;
545 Auto Seg-Loss: Searching Metric Surrogates for Semantic Segmentation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Auto Seg-Loss is the first general framework for searching surrogate losses for mainstream semantic segmentation metrics.		 Hao Li; Chenxin Tao; Xizhou Zhu; Xiaogang Wang; Gao Huang; Jifeng Dai;
546 Differentiable Segmentation of Sequences
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an architecture for effective gradient-based learning of segmented models for sequential data.		 Erik Scharw�chter; Jonathan Lennartz; Emmanuel M�ller;
547 Grounding Physical Object and Event Concepts Through Dynamic Visual Reasoning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a neural-symbolic framework to learn physical object and event concepts for video causal reasoning.		 Zhenfang Chen; Jiayuan Mao; Jiajun Wu; Kwan-Yee Kenneth Wong; Joshua B. Tenenbaum; Chuang Gan;
548 Learning Deep Features in Instrumental Variable Regression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Propose a novel deep learning based method for instrumental variable regression		 Liyuan Xu; Yutian Chen; Siddarth Srinivasan; Nando de Freitas; Arnaud Doucet; Arthur Gretton;
549 Online Adversarial Purification Based on Self-supervised Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we combine canonical supervised learning with self-supervised representation learning, and present Self-supervised Online Adversarial Purification (SOAP), a novel defense strategy that uses a self-supervised loss to purify adversarial examples at test-time.		 Changhao Shi; Chester Holtz; Gal Mishne;
550 Graph Information Bottleneck for Subgraph Recognition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a framework of Graph Information Bottleneck (GIB) for the subgraph recognition problem in deep graph learning.		 Junchi Yu; Tingyang Xu; Yu Rong; Yatao Bian; Junzhou Huang; Ran He;
551 In Search of Lost Domain Generalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our ERM baseline achieves state-of-the-art performance across many domain generalization benchmarks		 Ishaan Gulrajani; David Lopez-Paz;
552 Robust Curriculum Learning: from Clean Label Detection to Noisy Label Self-correction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: RoCL improves noisy label learning by periodical transitions from supervised learning of clean labeled data to self-supervision of wrongly-labeled data, where the data are selected according to training dynamics.		 Tianyi Zhou; Shengjie Wang; Jeff Bilmes;
553 Cross-Attentional Audio-Visual Fusion for Weakly-Supervised Action Localization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An audio-visual fusion technique, called multi-stage cross-attention, is developed to exploit the multi-modal representation in weakly-supervised action or event localization in untrimmed videos.		 Jun-Tae Lee; Mihir Jain; Hyoungwoo Park; Sungrack Yun;
554 CoCon: A Self-Supervised Approach for Controlled Text Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose CoCon to control the content of text generation from LMs by conditioning on content inputs at an interleave layer.		 Alvin Chan; Yew-Soon Ong; Bill Pung; Aston Zhang; Jie Fu;
555 Group Equivariant Generative Adversarial Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Equivariance to symmetry groups improves the generative adversarial synthesis of globally-symmetric images.		 Neel Dey; Antong Chen; Soheil Ghafurian;
556 What They Do When in Doubt: A Study of Inductive Biases in Seq2seq Learners
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Standard seq2seq models can infer perfectly different rules when presented with as little as one training example, showing strikingly different inductive biases than can be studied via description length		 Eugene Kharitonov; Rahma Chaabouni;
557 A Teacher-student Framework to Distill Future Trajectories
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We explore meta-learning a teacher network to efficiently incorporate privileged information such as trajectories.		 Alexander Neitz; Giambattista Parascandolo; Bernhard Sch�lkopf;
558 Learning A Latent Search Space for Routing Problems Using Variational Autoencoders
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use conditional variational autoencoders to learn continuous search spaces for routing problems that can be searched with any unconstrained continuous optimizer.		 Andr� Hottung; Bhanu Bhandari; Kevin Tierney;
559 Universal Approximation Power of Deep Residual Neural Networks Via Nonlinear Control Theory
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Using nonlinear control theory, it is shown that deep residual neural networks have the power of universal approximation with respect to the supremum norm.		 Paulo Tabuada; Bahman Gharesifard;
560 On The Universality of Rotation Equivariant Point Cloud Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We provide sufficient conditions for universality of rotation equivariant point cloud networks and use these conditions to show that current models are universal as well as for devising new universal architectures.		 Nadav Dym; Haggai Maron;
561 CT-Net: Channel Tensorization Network for Video Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To achieve convolution efficiency and feature-interaction sufficiency, we propose a Channel Tensorization Network (CT-Net), by treating the channel dimension of input feature as a multiplication of K sub-dimensions.		 Kunchang Li; Xianhang Li; Yali Wang; Jun Wang; Yu Qiao;
562 Learning to Live with Dale’s Principle: ANNs with Separate Excitatory and Inhibitory Units
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present two insights that enable DANNs to learn well: (1) DANNs are related to normalization schemes, and can be initialized such that the inhibition centres and standardizes the excitatory activity, (2) updates to inhibitory neuron parameters should be scaled using corrections based on the Fisher Information matrix.		 Jonathan Cornford; Damjan Kalajdzievski; Marco Leite; Am�lie Lamarquette; Dimitri Michael Kullmann; Blake Aaron Richards;
563 Uncertainty Estimation in Autoregressive Structured Prediction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A Deep Investigation of Ensemble-based Uncertainty Estimation for Autoregressive ASR and NMT models.		 Andrey Malinin; Mark Gales;
564 Transformer Protein Language Models Are Unsupervised Structure Learners
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Transformer attention maps directly represent protein contacts with state-of-the-art unsupervised precision.		 Roshan Rao; Joshua Meier; Tom Sercu; Sergey Ovchinnikov; Alexander Rives;
565 ANOCE: Analysis of Causal Effects with Multiple Mediators Via Constrained Structural Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Analysis of causal effects on the level of individual mediators via constrained structural learning, with application to the COVID-19 Spread in China.		 Hengrui Cai; Rui Song; Wenbin Lu;
566 Plan-Based Relaxed Reward Shaping for Goal-Directed Tasks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce Final-Volume-Preserving Reward Shaping, and show in a plan-based setting that it significantly increases the sample efficiency of reinforcement learning.		 Ingmar Schubert; Ozgur S Oguz; Marc Toussaint;
567 CcGAN: Continuous Conditional Generative Adversarial Networks for Image Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work proposes the continuous conditional generative adversarial network (CcGAN), the first generative model for image generation conditional on continuous, scalar conditions (termed as regression labels).		 Xin Ding; Yongwei Wang; Zuheng Xu; William J Welch; Z. Jane Wang;
568 Single-Photon Image Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Mathematical proof that the classical accuracy limit for single-photon image classification can be exceeded very substantially by employing a problem-tailored quantum transformation on the photon state.		 Thomas Fischbacher; Luciano Sbaiz;
569 Self-supervised Adversarial Robustness for The Low-label, High-data Regime
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper builds of Bootstrap Your Own Latents and proposes a self-supervised learning technique that can learn robust representations that are competitive with fully-supervised techniques.		 Sven Gowal; Po-Sen Huang; Aaron van den Oord; Timothy Mann; Pushmeet Kohli;
570 Uncertainty-aware Active Learning for Optimal Bayesian Classifier
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We focus on pool-based Bayesian active learning, for which we have proposed a new weighted MOCU method and analyzed its theoretical properties to demonstrate the better convergence properties than existing methods in this category.		 Guang Zhao; Edward Dougherty; Byung-Jun Yoon; Francis Alexander; Xiaoning Qian;
571 Skill Transfer Via Partially Amortized Hierarchical Planning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Partially amortized planning through hierarchy helps learn skills for complex control tasks		 Kevin Xie; Homanga Bharadhwaj; Danijar Hafner; Animesh Garg; Florian Shkurti;
572 Learning Continuous-time PDEs from Sparse Data with Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The paper introduces a method for learning partial differential equations on arbitrary spatial and temporal grids.		 Valerii Iakovlev; Markus Heinonen; Harri L�hdesm�ki;
573 Characterizing Signal Propagation to Close The Performance Gap in Unnormalized ResNets
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show how to train ResNets completely without normalization, and attain performance competitive with batch-normalized EfficientNets.		 Andrew Brock; Soham De; Samuel L Smith;
574 Robust Overfitting May Be Mitigated By Properly Learned Smoothening
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Mitigate robust overfitting by properly learned smoothening, establishing the new state-of-the-art bar in adversarial training		 Tianlong Chen; Zhenyu Zhang; Sijia Liu; Shiyu Chang; Zhangyang Wang;
575 Long Live The Lottery: The Existence of Winning Tickets in Lifelong Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Proposed novel bottom-up lifelong pruning effectively identify the winning tickets, which significantly improve the performance of learning over continual tasks		 Tianlong Chen; Zhenyu Zhang; Sijia Liu; Shiyu Chang; Zhangyang Wang;
576 Symmetry-Aware Actor-Critic for 3D Molecular Design
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Covariant actor-critic based on spherical harmonics that exploits symmetries to design molecules in 3D		 Gregor N. C. Simm; Robert Pinsler; G�bor Cs�nyi; Jos� Miguel Hern�ndez-Lobato;
577 PseudoSeg: Designing Pseudo Labels for Semantic Segmentation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper presents a new method that first demonstrates how well-calibrated soft pseudo labels obtained through wise fusion of predictions from diverse sources greatly improve consistency training for semantic segmentation.		 Yuliang Zou; Zizhao Zhang; Han Zhang; Chun-Liang Li; Xiao Bian; Jia-Bin Huang; Tomas Pfister;
578 NAS-Bench-ASR: Reproducible Neural Architecture Search for Speech Recognition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work we apply NAS for finding cell architecture for ASR models and release a comprehensive NAS-Bench dataset for reproducible NAS research.		 Abhinav Mehrotra; Alberto Gil C. P. Ramos; Sourav Bhattacharya; Lukasz Dudziak; Ravichander Vipperla; Thomas Chau; Mohamed S Abdelfattah; Samin Ishtiaq; Nicholas Donald Lane;
579 Scaling The Convex Barrier with Active Sets
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a specialised dual solver for a tight ReLU convex relaxation and show that it speeds up formal network verification.		 Alessandro De Palma; Harkirat Behl; Rudy R Bunel; Philip Torr; M. Pawan Kumar;
580 Local Convergence Analysis of Gradient Descent Ascent with Finite Timescale Separation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that there exists a range of finite learning ratios such that gradient descent-ascent converges to a critical point if and only if it is a strict local minmax equilibrium		 Tanner Fiez; Lillian J Ratliff;
581 Activation-level Uncertainty in Deep Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use 1D Gaussian Processes to introduce activation-level uncertainty in neural networks, which overcomes known limitations of (functional) Bayesian neural nets and obtains better results than the related deep Gaussian Processes.		 Pablo Morales-Alvarez; Daniel Hern�ndez-Lobato; Rafael Molina; Jos� Miguel Hern�ndez-Lobato;
582 Efficient Continual Learning with Modular Networks and Task-Driven Priors
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new benchmark allowing a detailed analysis of the properties of continual learning alogithms and a new modular neural network leveraging task-based priors to efficiently learn in the CL setting.		 Tom Veniat; Ludovic Denoyer; MarcAurelio Ranzato;
583 Amending Mistakes Post-hoc in Deep Networks By Leveraging Class Hierarchies
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Conditional risk framework exploiting the label hierarchy outperforms state of the art and makes a strong baseline for future explorations.		 Shyamgopal Karthik; Ameya Prabhu; Puneet K. Dokania; Vineet Gandhi;
584 Ringing ReLUs: Harmonic Distortion Analysis of Nonlinear Feedforward Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Nonlinearities create high-frequency distortions that affect network trainability.		 Christian H.X. Ali Mehmeti-G�pel; David Hartmann; Michael Wand;
585 Distance-Based Regularisation of Deep Networks for Fine-Tuning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We derive generalisation bounds applicable to fine-tuning, then demonstrate an algorithm that regularises these bounds improves fine-tuning performance.		 Henry Gouk; Timothy Hospedales; massimiliano pontil;
586 Genetic Soft Updates for Policy Evolution in Deep Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a novel mixed framework that combines the benefits of Evolutionary Algorithms and any DRL algorithms (including value-based ones); we support our claims on the beneficial policy improvement using recent formal verification tools.		 Enrico Marchesini; Davide Corsi; Alessandro Farinelli;
587 Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A computational-efficient GAN for few-shot hi-fi image dataset (converge on single gpu with few hours’ training, on 1024 resolution sub-hundred images).		 Bingchen Liu; Yizhe Zhu; Kunpeng Song; Ahmed Elgammal;
588 IsarStep: A Benchmark for High-level Mathematical Reasoning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a benchmark for high-level mathematical reasoning and study the reasoning capabilities of neural sequence-to-sequence models. We build a non-synthetic dataset from the largest repository of proofs written by human experts in a theorem prover.		 Wenda Li; Lei Yu; Yuhuai Wu; Lawrence C. Paulson;
589 Multi-Prize Lottery Ticket Hypothesis: Finding Accurate Binary Neural Networks By Pruning A Randomly Weighted Network
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new paradigm for learning compact yet accurate binary neural networks by pruning and quantizing randomly weighted full precision DNNs		 James Diffenderfer; Bhavya Kailkhura;
590 Average-case Acceleration for Bilinear Games and Normal Matrices
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We extend the framework of average-case optimal first-order methods to problems with non-symmetric matrices, which naturally arise in equilibrium finding for games.		 Carles Domingo-Enrich; Fabian Pedregosa; Damien Scieur;
591 ECONOMIC HYPERPARAMETER OPTIMIZATION WITH BLENDED SEARCH STRATEGY
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A low-cost hyperparameter optimization solution by blending global and local search methods with cost-based prioritization		 Chi Wang; Qingyun Wu; Silu Huang; Amin Saied;
592 BSQ: Exploring Bit-Level Sparsity for Mixed-Precision Neural Network Quantization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose bit-level sparsity inducing regularizer to induce mixed-percision quantization scheme in DNN with gradient-based training.		 Huanrui Yang; Lin Duan; Yiran Chen; Hai Li;
593 AutoLRS: Automatic Learning-Rate Schedule By Bayesian Optimization on The Fly
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we consider the question: Can we automatically tune the LR over the course of training without human involvement?		 Yuchen Jin; Tianyi Zhou; Liangyu Zhao; Yibo Zhu; Chuanxiong Guo; Marco Canini; Arvind Krishnamurthy;
594 BERTology Meets Biology: Interpreting Attention in Protein Language Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We analyze the internal representations of protein language models, and show that attention targets structural and functional properties of protein sequences.		 Jesse Vig; Ali Madani; Lav R. Varshney; Caiming Xiong; richard socher; Nazneen Rajani;
595 Learning Task-General Representations with Generative Neuro-Symbolic Modeling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: few-shot concept learning with generative neuro-symbolic models		 Reuben Feinman; Brenden M. Lake;
596 Zero-shot Synthesis with Group-Supervised Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To aid neural networks to envision objects with different attributes, we propose GSL which allows us to decompose inputs into a disentangled representation with swappable components, that can be recombined to synthesize new samples.		 Yunhao Ge; Sami Abu-El-Haija; Gan Xin; Laurent Itti;
597 Selective Classification Can Magnify Disparities Across Groups
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we find that while selective classification can improve average accuracies, it can simultaneously magnify existing accuracy disparities between various groups within a population, especially in the presence of spurious correlations.		 Erik Jones; Shiori Sagawa; Pang Wei Koh; Ananya Kumar; Percy Liang;
598 Better Fine-Tuning By Reducing Representational Collapse
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a lightweight augmentation to standard fine-tuning which outperforms previous methods across the board (i.e. SOTA on 3 summarization tasks, XNLI, RoBERTa on GLUE) while being computationally cheaper than other fine-tuning approaches.		 Armen Aghajanyan; Akshat Shrivastava; Anchit Gupta; Naman Goyal; Luke Zettlemoyer; Sonal Gupta;
599 Training Independent Subnetworks for Robust Prediction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that a deep neural network can be trained to give multiple independent predictions simultaneously, which results in a computationally efficient ensemble model.		 Marton Havasi; Rodolphe Jenatton; Stanislav Fort; Jeremiah Zhe Liu; Jasper Snoek; Balaji Lakshminarayanan; Andrew Mingbo Dai; Dustin Tran;
600 Meta-Learning of Structured Task Distributions in Humans and Machines
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We developed a novel meta-learning task with a structured task distribution and statistically equivalent null task distribution to show humans are more adept at the former whereas current meta-learning agents are more adept at the latter.		 Sreejan Kumar; Ishita Dasgupta; Jonathan Cohen; Nathaniel Daw; Thomas Griffiths;
601 BiPointNet: Binary Neural Network for Point Clouds
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present BiPointNet, the first model binarization approach to efficient deep learning on point clouds, targeting at extreme compression and acceleration.		 Haotong Qin; Zhongang Cai; Mingyuan Zhang; Yifu Ding; Haiyu Zhao; Shuai Yi; Xianglong Liu; Hao Su;
602 Benchmarks for Deep Off-Policy Evaluation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A benchmark proposal for off-policy evaluation and policy selection.		 Justin Fu; Mohammad Norouzi; Ofir Nachum; George Tucker; ziyu wang; Alexander Novikov; Mengjiao Yang; Michael R Zhang; Yutian Chen; Aviral Kumar; Cosmin Paduraru; Sergey Levine; Thomas Paine;
603 Planning from Pixels Using Inverse Dynamics Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: GLAMOR learns a latent world model by learning to predict action sequences conditioned on task completion.		 Keiran Paster; Sheila A. McIlraith; Jimmy Ba;
604 Understanding The Effects of Data Parallelism and Sparsity on Neural Network Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We accurately measure the effects of data parallelism and sparsity on neural network training and develop a theoretical analysis to precisely account for their effects.		 Namhoon Lee; Thalaiyasingam Ajanthan; Philip Torr; Martin Jaggi;
605 NOVAS: Non-convex Optimization Via Adaptive Stochastic Search for End-to-end Learning and Control
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work we propose the use of adaptive stochastic search as a building block for general, non-convex optimization operations within deep neural network architectures.		 Ioannis Exarchos; Marcus Aloysius Pereira; Ziyi Wang; Evangelos Theodorou;
606 MoVie: Revisiting Modulated Convolutions for Visual Counting and Beyond
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: 2020 VQA challenge winner; state-of-the-art performance on three counting benchmarks; can work beyond counting towards general visual reasoning		 Duy Kien Nguyen; Vedanuj Goswami; Xinlei Chen;
607 NeMo: Neural Mesh Models of Contrastive Features for Robust 3D Pose Estimation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce NeMo, a rendering-based approach to 3D pose estimation that models objects in terms of neural feature activations, instead of image intensities.		 Angtian Wang; Adam Kortylewski; Alan Yuille;
608 On Graph Neural Networks Versus Graph-Augmented MLPs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We establish a separation in representation power between GNNs and Graph-Augmented MLPs.		 Zhengdao Chen; Lei Chen; Joan Bruna;
609 Dual-mode ASR: Unify and Improve Streaming ASR with Full-context Modeling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Dual-mode ASR unifies and improves Streaming ASR with full-context modeling, simplifying the development and deployment workflow and improving both latency and accuracy.		 Jiahui Yu; Wei Han; Anmol Gulati; Chung-Cheng Chiu; Bo Li; Tara N Sainath; Yonghui Wu; Ruoming Pang;
610 Deep Learning Meets Projective Clustering
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We suggest a novel technique for compressing a fully connected layer (or an embedding layer).		 Alaa Maalouf; Harry Lang; Daniela Rus; Dan Feldman;
611 Reinforcement Learning with Random Delays
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a framework for Reinforcement Learning with random action and observation delays.		 Yann Bouteiller; Simon Ramstedt; Giovanni Beltrame; Christopher Pal; Jonathan Binas;
612 Isotropy in The Contextual Embedding Space: Clusters and Manifolds
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper reveals isotropy in the clustered contextual embedding space, and found low-dimensional manifolds in there.		 Xingyu Cai; Jiaji Huang; Yuchen Bian; Kenneth Church;
613 Spatio-Temporal Graph Scattering Transform
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We put forth a novel mathematically designed framework ST-GST to analyze spatio-temporal data.		 Chao Pan; Siheng Chen; Antonio Ortega;
614 Deployment-Efficient Reinforcement Learning Via Model-Based Offline Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel method that achieves both high sample-efficiency in offline RL and deployment-efficiency in online RL.		 Tatsuya Matsushima; Hiroki Furuta; Yutaka Matsuo; Ofir Nachum; Shixiang Gu;
615 GradSim: Differentiable Simulation for System Identification and Visuomotor Control
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Differentiable models of time-varying dynamics and image formation pipelines result in highly accurate physical parameter estimation from video		 J. Krishna Murthy; Miles Macklin; Florian Golemo; Vikram Voleti; Linda Petrini; Martin Weiss; Breandan Considine; J�r�me Parent-L�vesque; Kevin Xie; Kenny Erleben; Liam Paull; Florian Shkurti; Derek Nowrouzezahrai; Sanja Fidler;
616 Evaluations and Methods for Explanation Through Robustness Analysis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a suite of objective measurements for evaluating feature based explanations by the notion of robustness analysis; we further derive new explanation that captures different characteristics of explanation comparing to existing methods.		 Cheng-Yu Hsieh; Chih-Kuan Yeh; Xuanqing Liu; Pradeep Kumar Ravikumar; Seungyeon Kim; Sanjiv Kumar; Cho-Jui Hsieh;
617 RNNLogic: Learning Logic Rules for Reasoning on Knowledge Graphs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Learn Logic Rules for Reasoning on Knowledge Graphs.		 Meng Qu; Junkun Chen; Louis-Pascal Xhonneux; Yoshua Bengio; Jian Tang;
618 Can A Fruit Fly Learn Word Embeddings?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that a network motif from fruit fly brain can learn word embeddings.		 Yuchen Liang; Chaitanya Ryali; Benjamin Hoover; Saket Navlakha; Leopold Grinberg; Mohammed J Zaki; Dmitry Krotov;
619 Neural Representation and Generation for RNA Secondary Structures
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We investigate a new direction in computational drug discovery for designing large scale and complex macromolecular structures known as the RNAs.		 Zichao Yan; William L. Hamilton; Mathieu Blanchette;
620 WaNet – Imperceptible Warping-based Backdoor Attack
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an imperceptible backdoor attack based on image-warping, which can surpass both human and machine inspections.		 Tuan Anh Nguyen; Anh Tuan Tran;
621 LowKey: Leveraging Adversarial Attacks to Protect Social Media Users from Facial Recognition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We leverage adversarial attacks in our tool, LowKey, which protects social media users from invasive mass surveillance systems.		 Valeriia Cherepanova; Micah Goldblum; Harrison Foley; Shiyuan Duan; John P Dickerson; Gavin Taylor; Tom Goldstein;
622 Learning from Others’ Mistakes: Avoiding Dataset Biases Without Modeling Them
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Reducing a model’s reliance on dataset biases by encouraging a robust model to learn from a weak learner’s mistakes.		 Victor Sanh; Thomas Wolf; Yonatan Belinkov; Alexander M Rush;
623 Prototypical Contrastive Learning of Unsupervised Representations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an unsupervised representation learning method that bridges contrastive learning with clustering in an EM framework.		 Junnan Li; Pan Zhou; Caiming Xiong; Steven Hoi;
624 Extreme Memorization Via Scale of Initialization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Through studying the effect of scale of initialization on generalization, we come up with an alignment measure that correlations with generalization of deep models.		 Harsh Mehta; Ashok Cutkosky; Behnam Neyshabur;
625 Interactive Weak Supervision: Learning Useful Heuristics for Data Labeling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a framework and method for training classifiers on datasets without ground truth annotation by interacting with domain experts to discover good weak supervision sources.		 Benedikt Boecking; Willie Neiswanger; Eric Xing; Artur Dubrawski;
626 Adaptive Procedural Task Generation for Hard-Exploration Problems
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a framework which creates tasks as curricula via procedural generation to expedite reinforcement learning in hard-exploration problems.		 Kuan Fang; Yuke Zhu; Silvio Savarese; Fei-Fei Li;
627 Multi-timescale Representation Learning in LSTM Language Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work presents a theoretically-motivated analysis of memory and timescale in LSTM language models.		 Shivangi Mahto; Vy Ai Vo; Javier S. Turek; Alexander Huth;
628 Deep Encoder, Shallow Decoder: Reevaluating Non-autoregressive Machine Translation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that the speed disadvantage for autoregressive baselines to evaluate non-autoregressive machine translation is overestimated in three aspects: suboptimal layer allocation, insufficient speed measurement, and lack of knowledge distillation.		 Jungo Kasai; Nikolaos Pappas; Hao Peng; James Cross; Noah Smith;
629 Learning Better Structured Representations Using Low-rank Adaptive Label Smoothing
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an extension of label smoothing which improves generalization performance by adapting to the structure present in label space of structured prediction tasks.		 Asish Ghoshal; Xilun Chen; Sonal Gupta; Luke Zettlemoyer; Yashar Mehdad;
630 Information-theoretic Probing Explains Reliance on Spurious Features
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We find that feature extractability, measured by probing classifiers, can be viewed as an inductive bias: the more extractable a feature is after pre-training, the less statistical evidence needed during fine-tuning for the model to use the feature.		 Charles Lovering; Rohan Jha; Tal Linzen; Ellie Pavlick;
631 Molecule Optimization By Explainable Evolution
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel EM-like evolution-by-explanation algorithm alternating between an explainable graph model and a conditional generative model for molecule optimization.		 Binghong Chen; Tianzhe Wang; Chengtao Li; Hanjun Dai; Le Song;
632 Anchor & Transform: Learning Sparse Embeddings for Large Vocabularies
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: End-to-end learning of sparse embeddings for large vocabularies with a Bayesian nonparametric interpretation that results in up to 40x smaller embedding tables.		 Paul Pu Liang; Manzil Zaheer; Yuan Wang; Amr Ahmed;
633 PSTNet: Point Spatio-Temporal Convolution on Point Cloud Sequences
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper proposes a point spatio-temporal (PST) convolution to learn representations of raw point cloud sequences.		 Hehe Fan; Xin Yu; Yuhang Ding; Yi Yang; Mohan Kankanhalli;
634 Group Equivariant Conditional Neural Processes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A model for regression that learns conditional distributions of a stochastic process, by incorporating group equivariance into Conditional Neural Processes.		 Makoto Kawano; Wataru Kumagai; Akiyoshi Sannai; Yusuke Iwasawa; Yutaka Matsuo;
635 When Does Preconditioning Help Or Hurt Generalization?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Characterized the population risk of preconditioned least squares regression in the overparameterized regime and determined the optimal preconditioner for generalization.		 Shun-ichi Amari; Jimmy Ba; Roger Baker Grosse; Xuechen Li; Atsushi Nitanda; Taiji Suzuki; Denny Wu; Ji Xu;
636 Learning Reasoning Paths Over Semantic Graphs for Video-grounded Dialogues
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a novel approach to learn reasoning paths over semantic graphs which are built upon dialogue context at each turn, for video-grounded dialogues.		 Hung Le; Nancy F. Chen; Steven Hoi;
637 Prototypical Representation Learning for Relation Extraction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Instead of solely relying on the supervision from labels (which could be noisy), we propose to infer a latent prototype for each relation from contextual information to best explore the intrinsic semantics of relations.		 Ning Ding; Xiaobin Wang; Yao Fu; Guangwei Xu; Rui Wang; Pengjun Xie; Ying Shen; Fei Huang; Hai-Tao Zheng; Rui Zhang;
638 Layer-adaptive Sparsity for The Magnitude-based Pruning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose LAMP, a general-purpose layerwise sparsity selection scheme for magnitude pruning.		 Jaeho Lee; Sejun Park; Sangwoo Mo; Sungsoo Ahn; Jinwoo Shin;
639 Refining Deep Generative Models Via Wasserstein Gradient Flows
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A method of refining samples from deep generative models using the discriminator gradient flow of f-divergences.		 Abdul Fatir Ansari; Ming Liang Ang; Harold Soh;
640 Explaining The Efficacy of Counterfactually Augmented Data
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a framework for thinking about counterfactually augmented data and make strides towards understanding its benefits in out-of-domain generalization.		 Divyansh Kaushik; Amrith Setlur; Eduard H Hovy; Zachary Chase Lipton;
641 Lipschitz Recurrent Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a provably stable parameterization for continuous-time Lipschitz Recurrent Neural Networks that can employ high order integration schemes and outperform existing RNNs in performance, robustness, and conditioning.		 N. Benjamin Erichson; Omri Azencot; Alejandro Queiruga; Liam Hodgkinson; Michael W. Mahoney;
642 Learning Hyperbolic Representations of Topological Features
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a method to learn representations of topological features (i.e., persistence diagrams) on hyperbolic spaces.		 Panagiotis Kyriakis; Iordanis Fostiropoulos; Paul Bogdan;
643 Bidirectional Variational Inference for Non-Autoregressive Text-to-Speech
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, a novel non-autoregressive text-to-speech model based on bidirectional-inference variational autoencoder called BVAE-TTS is proposed.		 Yoonhyung Lee; Joongbo Shin; Kyomin Jung;
644 Risk-Averse Offline Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the first risk-averse reinforcement learning algorithm in the fully offline setting.		 N�ria Armengol Urp�; Sebastian Curi; Andreas Krause;
645 Group Equivariant Stand-Alone Self-Attention For Vision
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We provide a general self-attention formulation to impose group equivariance to arbitrary symmetry groups.		 David W. Romero; Jean-Baptiste Cordonnier;
646 A Better Alternative to Error Feedback for Communication-Efficient Distributed Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a new and theoretically and practically better alternative to EF for dealing with contractive compressors.		 Samuel Horv�th; Peter Richtarik;
647 Neural Delay Differential Equations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new class of continuous-depth neural networks with delay, named as Neural Delay Differential Equations and having better representation capability outperforming the Neural ODEs.		 Qunxi Zhu; Yao Guo; Wei Lin;
648 Capturing Label Characteristics in VAEs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a principled approach to incorporating labels in VAEs that captures the rich characteristic information associated with those labels.		 Tom Joy; Sebastian Schmon; Philip Torr; Siddharth N; Tom Rainforth;
649 Graph Edit Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that graph neural networks can predict graph edits and are connected to the graph edit distance via graph mappings		 Benjamin Paassen; Daniele Grattarola; Daniele Zambon; Cesare Alippi; Barbara Eva Hammer;
650 InfoBERT: Improving Robustness of Language Models from An Information Theoretic Perspective
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel learning framework, InfoBERT, for robust fine-tuning of pre-trained language models from an information-theoretic perspective, and achieve state-of-the-art robust accuracy over several adversarial datasets on NLI and QA tasks.		 Boxin Wang; Shuohang Wang; Yu Cheng; Zhe Gan; Ruoxi Jia; Bo Li; Jingjing Liu;
651 DrNAS: Dirichlet Neural Architecture Search
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper proposes a novel differentiable architecture search method by formulating it into a distribution learning problem.		 Xiangning Chen; Ruochen Wang; Minhao Cheng; Xiaocheng Tang; Cho-Jui Hsieh;
652 Drop-Bottleneck: Learning Discrete Compressed Representation for Noise-Robust Exploration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our novel IB method, Drop-Bottleneck, discretely drops task-irrelevant input features to build the compressed representation and shows state-of-the-art performance on noisy, sparse-reward navigation tasks in reinforcement learning.		 Jaekyeom Kim; Minjung Kim; Dongyeon Woo; Gunhee Kim;
653 Monte-Carlo Planning and Learning with Language Action Value Estimates
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present Monte-Carlo planning with Language Action Value Estimates (MC-LAVE) that combines a Monte-Carlo tree search with language-driven exploration for Interactive Fiction games.		 Youngsoo Jang; Seokin Seo; Jongmin Lee; Kee-Eung Kim;
654 Robust Early-learning: Hindering The Memorization of Noisy Labels
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, motivated by the \textit{lottery ticket hypothesis} which shows that only partial parameters are important for generalization, we find that only partial parameters are important for fitting clean labels and generalize well, which we term as \textit{critical parameters}; while the other parameters tend to fit noisy labels and cannot generalize well, which we term as \textit{non-critical parameters}.		 Xiaobo Xia; Tongliang Liu; Bo Han; Chen Gong; Nannan Wang; Zongyuan Ge; Yi Chang;
655 Identifying Physical Law of Hamiltonian Systems Via Meta-Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce meta-learning algorithms to identify the shared representation of Hamiltonian systems.		 Seungjun Lee; Haesang Yang; Woojae Seong;
656 Reweighting Augmented Samples By Minimizing The Maximal Expected Loss
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: a new reweighting strategy on augmented samples		 Mingyang Yi; Lu Hou; Lifeng Shang; Xin Jiang; Qun Liu; Zhi-Ming Ma;
657 Understanding and Improving Encoder Layer Fusion in Sequence-to-Sequence Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, our main contribution is to take a step further in understanding EncoderFusion.		 Xuebo Liu; Longyue Wang; Derek F. Wong; Liang Ding; Lidia S. Chao; Zhaopeng Tu;
658 Spatial Dependency Networks: Neural Layers for Improved Generative Image Modeling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel neural network layer for improved generative modeling of images, applied to variational autoencoders.		 �orde Miladinovic; Aleksandar Stanic; Stefan Bauer; J�rgen Schmidhuber; Joachim M. Buhmann;
659 Deep Repulsive Clustering of Ordered Data Based on Order-Identity Decomposition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A deep clustering algorithm for ordered data is proposed based on the order-identity decomposition.		 Seon-Ho Lee; Chang-Su Kim;
660 Revisiting Locally Supervised Learning: An Alternative to End-to-end Training
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We provide a deep understanding of locally supervised learning, and make it perform on par with end-to-end training, while with significantly reduced GPUs memory footprint.		 Yulin Wang; Zanlin Ni; Shiji Song; Le Yang; Gao Huang;
661 Nonseparable Symplectic Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To solve the problem, we propose a novel neural network architecture, Nonseparable Symplectic Neural Networks (NSSNNs), to uncover and embed the symplectic structure of a nonseparable Hamiltonian system from limited observation data.		 Shiying Xiong; Yunjin Tong; Xingzhe He; Cheng Yang; Shuqi Yang; Bo Zhu;
662 Gradient Origin Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new model that uses the negative gradient of the loss with respect to the origin as a latent vector is found to be superior to equivalent networks.		 Sam Bond-Taylor; Chris G. Willcocks;
663 Learning to Sample with Local and Global Contexts in Experience Replay Buffer
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a learning-based neural replay which calculates the relative importance to sample experience for off-policy RL.		 Youngmin Oh; Kimin Lee; Jinwoo Shin; Eunho Yang; Sung Ju Hwang;
664 Provable Rich Observation Reinforcement Learning with Combinatorial Latent States
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a problem setup and a provable reinforcement learning algorithm for rich-observation problems with latent combinatorially large state space.		 Dipendra Misra; Qinghua Liu; Chi Jin; John Langford;
665 Sharper Generalization Bounds for Learning with Gradient-dominated Objective Functions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop sharper generalization bounds for learning with gradient-dominated objective functions.		 Yunwen Lei; Yiming Ying;
666 Rapid Neural Architecture Search By Learning to Generate Graphs from Datasets
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an efficient NAS framework that is trained once on a database consisting of datasets and pretrained networks and can rapidly generate a neural architecture for a novel dataset.		 Hayeon Lee; Eunyoung Hyung; Sung Ju Hwang;
667 Relating By Contrasting: A Data-efficient Framework for Multimodal Generative Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To mitigate this, we develop a novel contrastive framework for generative model learning, allowing us to train the model not just by the commonality between modalities, but by the distinction between related and unrelated multimodal data.		 Yuge Shi; Brooks Paige; Philip Torr; Siddharth N;
668 FedMix: Approximation of Mixup Under Mean Augmented Federated Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a new federated framework, Mean Augmented Federated Learning (MAFL), and propose an efficient algorithm, Federated Mixup (FedMix), which shows good performance on difficult non-iid situations.		 Tehrim Yoon; Sumin Shin; Sung Ju Hwang; Eunho Yang;
669 Generalized Variational Continual Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We generalize VCL and Online-EWC and combine with task-specific FiLM layers		 Noel Loo; Siddharth Swaroop; Richard E Turner;
670 Understanding and Improving Lexical Choice in Non-Autoregressive Translation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We reveal the side effect of knowledge distillation from lexical choice perspective for Non-autoregressvie machine translation, and then propose a simple yet effective approach to improve it.		 Liang Ding; Longyue Wang; Xuebo Liu; Derek F. Wong; Dacheng Tao; Zhaopeng Tu;
671 Bayesian Context Aggregation for Neural Processes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a Bayesian Aggregation mechanism for Neural Process-based models which improves upon traditional mean aggregation.		 Michael Volpp; Fabian Fl�renbrock; Lukas Grossberger; Christian Daniel; Gerhard Neumann;
672 Variational Intrinsic Control Revisited
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Revisitation of Variational Intrinsic Control (VIC) for the optimal behavior of implicit VIC under stochastic dynamics.		 Taehwan Kwon;
673 Implicit Gradient Regularization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We have found a hidden form of regularization in gradient descent – Implicit Gradient Regularization – that biases overparameterized models towards flat, low test error solutions and helps us to understand why deep learning works so well.		 David Barrett; Benoit Dherin;
674 Return-Based Contrastive Representation Learning for Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel contrastive learning based auxiliary task which forces the learnt representations to discriminate state-action pairs with different returns and achieve superior performance on complex tasks in terms of sample effiency.		 Guoqing Liu; Chuheng Zhang; Li Zhao; Tao Qin; Jinhua Zhu; Li Jian; Nenghai Yu; Tie-Yan Liu;
675 Scalable Bayesian Inverse Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A variational inference approach to Bayesian inverse reinforcement learning.		 Alex James Chan; Mihaela van der Schaar;
676 Exemplary Natural Images Explain CNN Activations Better Than State-of-the-Art Feature Visualization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Using human psychophysical experiments, we show that natural images can be significantly more informative for interpreting neural network activations than synthetic feature visualizations.		 Judy Borowski; Roland Simon Zimmermann; Judith Schepers; Robert Geirhos; Thomas S. A. Wallis; Matthias Bethge; Wieland Brendel;
677 LiftPool: Bidirectional ConvNet Pooling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: By adopting the philosophy of the classical Lifting Scheme from signal processing, we propose LiftPool for bidirectional pooling layers, including LiftDownPool and LiftUpPool.		 Jiaojiao Zhao; Cees G. M. Snoek;
678 Adversarial Score Matching and Improved Sampling for Image Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Combining GANs with score matching and using Consistent Sampling (as an alternative to Langevin dynamics) for improved generative modeling		 Alexia Jolicoeur-Martineau; R�mi Pich�-Taillefer; Ioannis Mitliagkas; Remi Tachet des Combes;
679 Transient Non-stationarity and Generalisation in Deep Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We find that transient non-stationarity can worsen generalization in reinforcement learning and propose a method to overcome this effeect.		 Maximilian Igl; Gregory Farquhar; Jelena Luketina; Wendelin Boehmer; Shimon Whiteson;
680 On The Origin of Implicit Regularization in Stochastic Gradient Descent
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: For small learning rates, the iterates of SGD stay close to the path of gradient flow on a modified loss function containing an implicit regularizer.		 Samuel L Smith; Benoit Dherin; David Barrett; Soham De;
681 Analyzing The Expressive Power of Graph Neural Networks in A Spectral Perspective
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper aims to analyse of the expressive power of Graph Neural Network in spectral domain.		 Muhammet Balcilar; Guillaume Renton; Pierre H�roux; Benoit Ga�z�re; S�bastien Adam; Paul Honeine;
682 Pruning Neural Networks at Initialization: Why Are We Missing The Mark?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Methods for pruning neural nets at initialization perform the same or better when shuffling or reinitializing the weights they prune in each layer, a way in which they differ from SOTA weight-pruning methods after training.		 Jonathan Frankle; Gintare Karolina Dziugaite; Daniel Roy; Michael Carbin;
683 SkipW: Resource Adaptable RNN with Strict Upper Computational Limit
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Skip-Window is a method to allow recurrent neural networks (RNNs) to trade off accuracy for computational cost during the analysis of a sequence while keeping a strict upper computational limit.		 Tsiry Mayet; Anne Lambert; Pascal Leguyadec; Francoise Le Bolzer; Fran�ois Schnitzler;
684 Into The Wild with AudioScope: Unsupervised Audio-Visual Separation of On-Screen Sounds
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an open-domain unsupervised audio-visual on-screen separation system trained and tested on in-the-wild videos.		 Efthymios Tzinis; Scott Wisdom; Aren Jansen; Shawn Hershey; Tal Remez; Dan Ellis; John R. Hershey;
685 Simple Augmentation Goes A Long Way: ADRL for DNN Quantization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Augments the neural networks in Deep Reinforcement Learning(DRL) with a complementary scheme to boost the performance of learning and solve the common low convergence problem in the early stage of DRL		 Lin Ning; Guoyang Chen; Weifeng Zhang; Xipeng Shen;
686 Few-Shot Bayesian Optimization with Deep Kernel Surrogates
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Model-agnostic meta-learning meets Bayesian optimization to speed-up hyperparameter optimization by learning on metadata from different data sets.		 Martin Wistuba; Josif Grabocka;
687 AdaSpeech: Adaptive Text to Speech for Custom Voice
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose AdaSpeech, an adaptive TTS system for high-quality and efficient adaptation of new speaker in custom voice.		 Mingjian Chen; Xu Tan; Bohan Li; Yanqing Liu; Tao Qin; sheng zhao; Tie-Yan Liu;
688 HeteroFL: Computation and Communication Efficient Federated Learning for Heterogeneous Clients
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose a new federated learning framework named HeteroFL to train heterogeneous local models with varying computation complexities.		 Enmao Diao; Jie Ding; Vahid Tarokh;
689 DINO: A Conditional Energy-Based GAN for Domain Translation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A framework for domain translation which uses a novel mechanism for conditioning Energy-based GANs.		 Konstantinos Vougioukas; Stavros Petridis; Maja Pantic;
690 Universal Weakly Supervised Segmentation By Pixel-to-Segment Contrastive Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a unified pixel-to-segment contrastive learning loss formulation for weakly supervised semantic segmentation with various types of annotations.		 Tsung-Wei Ke; Jyh-Jing Hwang; Stella Yu;
691 PC2WF: 3D Wireframe Reconstruction from Raw Point Clouds
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An end-to-end trainable deep neural network for converting a 3D point cloud into a wireframe model.		 Yujia Liu; Stefano D’Aronco; Konrad Schindler; Jan Dirk Wegner;
692 Multi-resolution Modeling of A Discrete Stochastic Process Identifies Causes of Cancer
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We integrate a deep learning framework with a probabilistic model to learn a discrete stochastic process at arbitrary length scales, the method accurately and efficiently model mutations load in a tumor and detect cancer driver mutations genome-wide		 Adam Uri Yaari; Maxwell Sherman; Oliver Clarke Priebe; Po-Ru Loh; Boris Katz; Andrei Barbu; Bonnie Berger;
693 C-Learning: Horizon-Aware Cumulative Accessibility Estimation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce C-learning, a Q-learning inspired method to learn horizon-dependent policies for goal reaching.		 Panteha Naderian; Gabriel Loaiza-Ganem; Harry J. Braviner; Anthony L. Caterini; Jesse C. Cresswell; Tong Li; Animesh Garg;
694 Shapley Explanation Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: To enable new capabilities, we propose to use Shapley values as inter-layer representations in deep neural networks rather than as post-hoc explanations.		 Rui Wang; Xiaoqian Wang; David I. Inouye;
695 The Role of Disentanglement in Generalisation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Disentangled models do not achieve compositional generalization when tested systematically.		 Milton Llera Montero; Casimir JH Ludwig; Rui Ponte Costa; Gaurav Malhotra; Jeffrey Bowers;
696 Learning N:M Fine-grained Structured Sparse Neural Networks From Scratch
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: a simple yet universal recipe to learn N:M sparse neural networks from scratch		 Aojun Zhou; Yukun Ma; Junnan Zhu; Jianbo Liu; Zhijie Zhang; Kun Yuan; Wenxiu Sun; Hongsheng Li;
697 Unsupervised Meta-Learning Through Latent-Space Interpolation in Generative Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use interpolation in generative models latent space to generate tasks for unsupervised meta-learninig.		 Siavash Khodadadeh; Sharare Zehtabian; Saeed Vahidian; Weijia Wang; Bill Lin; Ladislau Boloni;
698 On Data-Augmentation and Consistency-Based Semi-Supervised Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a simple and natural framework leveraging the Hidden Manifold Model to study modern SSL methods.		 Atin Ghosh; Alexandre H. Thiery;
699 Learning from Demonstration with Weakly Supervised Disentanglement
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a generative model-based approach to learning interpretable robot trajectory representations from demonstrations (image embeddings and end-effector trajectories) paired with coarse labels, which provide a form of weak supervision.		 Yordan Hristov; Subramanian Ramamoorthy;
700 Neurally Augmented ALISTA
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce Neurally Augmented ALISTA, extending ALISTA to compute adaptive parameters to achieve improved recovery of individual sparse target vectors.		 Freya Behrens; Jonathan Sauder; Peter Jung;
701 Shape Or Texture: Understanding Discriminative Features in CNNs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Exploring and quantifying shape information encoded in CNNs.		 Md Amirul Islam; Matthew Kowal; Patrick Esser; Sen Jia; Bj�rn Ommer; Konstantinos G. Derpanis; Neil Bruce;
702 Convex Potential Flows: Universal Probability Distributions with Optimal Transport and Convex Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose to use an input-convex neural network to parameterize an invertible model with universal density approximation guarantees.		 Chin-Wei Huang; Ricky T. Q. Chen; Christos Tsirigotis; Aaron Courville;
703 Wasserstein Embedding for Graph Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Wasserstein Embedding for Graph Learning (WEGL) is a novel and fast framework for embedding entire graphs into a vector space in which the Euclidean distance between representations approximates the 2-Wasserstein distance.		 Soheil Kolouri; Navid Naderializadeh; Gustavo K. Rohde; Heiko Hoffmann;
704 Meta-learning with Negative Learning Rates
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show theoretically that the optimal inner learning rate of MAML during training is always negative in a family of models		 Alberto Bernacchia;
705 Representing Partial Programs with Blended Abstract Semantics
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use a combination of concrete execution and learned neural semantics to represent partial programs, resulting in more accurate program synthesis.		 Maxwell Nye; Yewen Pu; Matthew Bowers; Jacob Andreas; Joshua B. Tenenbaum; Armando Solar-Lezama;
706 Fast Convergence of Stochastic Subgradient Method Under Interpolation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper studies the behaviour of the stochastic subgradient descent (SSGD) method applied to over-parameterized nonsmooth optimization problems that satisfy an interpolation condition.		 Huang Fang; Zhenan Fan; Michael Friedlander;
707 A Hypergradient Approach to Robust Regression Without Correspondence
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a differentiable programming framework for the regression without correspondence problem.		 Yujia Xie; Yixiu Mao; Simiao Zuo; Hongteng Xu; Xiaojing Ye; Tuo Zhao; Hongyuan Zha;
708 On The Role of Planning in Model-based Deep Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An empirical investigation into how planning drives performance in model-based RL algorithms.		 Jessica B Hamrick; Abram L. Friesen; Feryal Behbahani; Arthur Guez; Fabio Viola; Sims Witherspoon; Thomas Anthony; Lars Holger Buesing; Petar Velickovic; Theophane Weber;
709 Trajectory Prediction Using Equivariant Continuous Convolution
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our model, ECCO, uses rotationally-equivariant continuous convolution to improve generalization in trajectory prediction.		 Robin Walters; Jinxi Li; Rose Yu;
710 Grounding Language to Autonomously-Acquired Skills Via Goal Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new RL architecture called Language-Goal-Behavior that proposes to decouple skill learning and language grounding via the introduction of an intermediate semantic goal representation.		 Ahmed Akakzia; C�dric Colas; Pierre-Yves Oudeyer; Mohamed CHETOUANI; Olivier Sigaud;
711 Chaos of Learning Beyond Zero-sum and Coordination Via Game Decompositions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We characterize games in which popular learning algorithms exhibit Lyapunov chaos.		 Yun Kuen Cheung; Yixin Tao;
712 Isometric Transformation Invariant and Equivariant Graph Convolutional Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We developed isometric transformation invariant and equivariant graph convolutional networks, which shows high prediction performance and computational efficiency.		 Masanobu Horie; Naoki Morita; Toshiaki Hishinuma; Yu Ihara; Naoto Mitsume;
713 R-GAP: Recursive Gradient Attack on Privacy
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: However, there is a fundamental lack of theoretical understanding of how and when gradients can lead to unique recovery of original data. Our research fills this gap by providing a closed-form recursive procedure to recover data from gradients in deep neural networks.		 Junyi Zhu; Matthew B. Blaschko;
714 Multi-Level Local SGD: Distributed SGD for Heterogeneous Hierarchical Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose Multi-Level Local SGD, a distributed stochastic gradient method for learning a smooth, non-convex objective in a multi-level communication network with heterogeneous workers.		 Timothy Castiglia; Anirban Das; Stacy Patterson;
715 GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper we demonstrate conditional computation as a remedy to the above mentioned impediments, and demonstrate its efficacy and utility.		 Dmitry Lepikhin; HyoukJoong Lee; Yuanzhong Xu; Dehao Chen; Orhan Firat; Yanping Huang; Maxim Krikun; Noam Shazeer; Zhifeng Chen;
716 Representation Learning for Improved Interpretability and Classification Accuracy of Clinical Factors from EEG
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use disentangled representations of EEG signals to improve performance on clinical classification tasks, provide interpretable recommendations for post-hoc analysis and allow for extraction of ERPs from novel single EEG trajectories.		 Garrett Honke; Irina Higgins; Nina Thigpen; Vladimir Miskovic; Katie Link; Sunny Duan; Pramod Gupta; Julia Klawohn; Greg Hajcak;
717 Multiplicative Filter Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose and empirically demonstrate that an arguably simpler class of function approximators can work just as well for such problems: multiplicative filter networks.		 Rizal Fathony; Anit Kumar Sahu; Devin Willmott; J Zico Kolter;
718 Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Masks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a method for analyzing emerging functional modularity in neural networks based on differentiable weight masks and use it to point out important issues in current-day neural networks.		 R�bert Csord�s; Sjoerd van Steenkiste; J�rgen Schmidhuber;
719 Modeling The Second Player in Distributionally Robust Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use generative neural models to define the uncertainty set in distributionally robust optimization, and show that this helps train more robust classifiers		 Paul Michel; Tatsunori Hashimoto; Graham Neubig;
720 Private Post-GAN Boosting
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose Private post-GAN boosting (Private PGB), a differentially private method that combines samples produced by the sequence of generators obtained during GAN training to create a high-quality synthetic dataset.		 Marcel Neunhoeffer; Steven Wu; Cynthia Dwork;
721 Flowtron: An Autoregressive Flow-based Generative Network for Text-to-Speech Synthesis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper we propose Flowtron: an autoregressive flow-based generative network for text-to-speech synthesis with style transfer and speech varation.		 Rafael Valle; Kevin J. Shih; Ryan Prenger; Bryan Catanzaro;
722 Learning Structural Edits Via Incremental Tree Transformations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A generic incremental editing model for tree-structured data		 Ziyu Yao; Frank F. Xu; Pengcheng Yin; Huan Sun; Graham Neubig;
723 Sample-Efficient Automated Deep Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: SEARL trains a population of off-policy RL agents while simultaneously optimizing the hyperparameters and the neural architecture sample-efficiently.		 J�rg K.H. Franke; Gregor Koehler; Andr� Biedenkapp; Frank Hutter;
724 Unsupervised Discovery of 3D Physical Objects
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an unsupervised framework for discovery 3D physical objects and show that these 3D objects to be used for tasks mimicking early infant cognition.		 Yilun Du; Kevin A. Smith; Tomer Ullman; Joshua B. Tenenbaum; Jiajun Wu;
725 Global Optimality of Softmax Policy Gradient with Single Hidden Layer Neural Networks in The Mean-field Regime
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove that softmax policy gradient algorithms with single hidden layer neural networks in the mean-field regime can be expressed as a gradient flow in Wasserstein space and prove that all the fixed points of such dynamics are global optimizers		 Andrea Agazzi; Jianfeng Lu;
726 Extracting Strong Policies for Robotics Tasks from Zero-order Trajectory Optimizers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an adaptively guided imitation learning method that is able to extract strong policies for hard robotic tasks from zero-order trajectory optimizers.		 Cristina Pinneri; Shambhuraj Sawant; Sebastian Blaes; Georg Martius;
727 Temporally-Extended E-Greedy Exploration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We discuss a new framework for option-based exploration, present a thorough empirical study of a simple, generally applicable set of options within this framework, and observe improved performance over state-of-the-art agents and exploration methods.		 Will Dabney; Georg Ostrovski; Andre Barreto;
728 Rapid Task-Solving in Novel Environments
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our agents meta-learn to explore, build models on-the-fly, and plan, enabling them to rapidly solve sequences of tasks in unfamiliar environments. To enable progress toward RTS, we introduce two challenge domains: (1) a minimal RTS challenge called the Memory\&Planning Game and (2) One-Shot StreetLearn Navigation, which introduces scale and complexity from real-world data.		 Samuel Ritter; Ryan Faulkner; Laurent Sartran; Adam Santoro; Matthew Botvinick; David Raposo;
729 Tradeoffs in Data Augmentation: An Empirical Study
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We quantify mechanisms of how data augmentation works with two metrics we introduce: Affinity and Diversity.		 Raphael Gontijo-Lopes; Sylvia Smullin; Ekin Dogus Cubuk; Ethan Dyer;
730 Multiscale Score Matching for Out-of-Distribution Detection
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Using score estimates at multiple noise scales outperforms state-of-the-art in out-of-distribution detection.		 Ahsan Mahmood; Junier Oliva; Martin Andreas Styner;
731 Understanding Over-parameterization in Generative Adversarial Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an analysis of over-paramterization in GANs both theoretically and empirically.		 Yogesh Balaji; Mohammadmahdi Sajedi; Neha Mukund Kalibhat; Mucong Ding; Dominik St�ger; Mahdi Soltanolkotabi; Soheil Feizi;
732 Go with The Flow: Adaptive Control for Neural ODEs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper presents a new method to enhance Neural ODEs representation power by dynamically controlling their weight parametrization.		 Mathieu Chalvidal; Matthew Ricci; Rufin VanRullen; Thomas Serre;
733 Linear Last-iterate Convergence in Constrained Saddle-point Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove Optimistic Gradient Descent Ascent (OGDA) and Optimistic Multiplicative Weights Update (OMWU) converge exponentially fast to the Nash equilibrium in the sense of last-iterate in various game settings including matrix games.		 Chen-Yu Wei; Chung-Wei Lee; Mengxiao Zhang; Haipeng Luo;
734 Learning Advanced Mathematical Computations from Examples
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We train transformers to predict qualitative and numerical properties of differential equations		 Francois Charton; Amaury Hayat; Guillaume Lample;
735 WaveGrad: Estimating Gradients for Waveform Generation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper introduces WaveGrad, a conditional model for waveform generation through estimating gradients of the data density.		 Nanxin Chen; Yu Zhang; Heiga Zen; Ron J Weiss; Mohammad Norouzi; William Chan;
736 SALD: Sign Agnostic Learning with Derivatives
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Sign agnostic learning with derivatives for learning high fidelity 3D implicit neural representations shape space from raw data.		 Matan Atzmon; Yaron Lipman;
737 Generalized Energy Based Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce the Generalized Energy Based Model (GEBM) for generative modelling.		 Michael Arbel; Liang Zhou; Arthur Gretton;
738 Long Range Arena : A Benchmark for Efficient Transformers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Better benchmarking for Xformers		 Yi Tay; Mostafa Dehghani; Samira Abnar; Yikang Shen; Dara Bahri; Philip Pham; Jinfeng Rao; Liu Yang; Sebastian Ruder; Donald Metzler;
739 Beyond Categorical Label Representations for Image Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the role of label representations for standard image classification task and found high-dimensional hign-entropy labes generally lead to more robust and data-efficient networks.		 Boyuan Chen; Yu Li; Sunand Raghupathi; Hod Lipson;
740 CoCo: Controllable Counterfactuals for Evaluating Dialogue State Trackers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: CoCo is a principled method to more flexibly evaluate the robustness of the DST component of TOD systems.		 SHIYANG LI; Semih Yavuz; Kazuma Hashimoto; Jia Li; Tong Niu; Nazneen Rajani; Xifeng Yan; Yingbo Zhou; Caiming Xiong;
741 Stochastic Security: Adversarial Defense Using Long-Run Dynamics of Energy-Based Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our defensive transformation using long-run MCMC sampling with a convergent EBM is the first method to successfully defend naturally-trained classifiers against adversarial attacks.		 Mitch Hill; Jonathan Craig Mitchell; Song-Chun Zhu;
742 X2T: Training An X-to-Text Typing Interface with Online Learning from User Feedback
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use online learning from user feedback to train an adaptive interface for typing words using inputs like eye gaze.		 Jensen Gao; Siddharth Reddy; Glen Berseth; Anca Dragan; Sergey Levine;
743 Mapping The Timescale Organization of Neural Language Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We demonstrated a model-free technique for mapping the timescale organization in neural network models, and we applied this method to reveal a hierarchical timescale organization within LSTM language models.		 Hsiang-Yun Sherry Chien; Jinhan Zhang; Christopher Honey;
744 PDE-Driven Spatiotemporal Disentanglement
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a novel interpretation of spatiotemporal disentanglement, inducing a simple and performant disentangled prediction model.		 J�r�mie Don�; Jean-Yves Franceschi; sylvain lamprier; patrick gallinari;
745 OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An effective way to leverage multimodal offline behavioral data is to extract a continuous space of primitives, and use it for downstream task learning.		 Anurag Ajay; Aviral Kumar; Pulkit Agrawal; Sergey Levine; Ofir Nachum;
746 Does Enhanced Shape Bias Improve Neural Network Robustness to Common Corruptions?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that robustness on common corruptions donot correlate with strong shape bias but with the effective data augmentation strategies like stylization		 Chaithanya Kumar Mummadi; Ranjitha Subramaniam; Robin Hutmacher; Julien Vitay; Volker Fischer; Jan Hendrik Metzen;
747 Directed Acyclic Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose DAGNN, a graph neural network tailored to directed acyclic graphs that outperforms conventional GNNs by leveraging the partial order as strong inductive bias besides other suitable architectural features.		 Veronika Thost; Jie Chen;
748 QPLEX: Duplex Dueling Multi-Agent Q-Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel multi-agent Q-learning algorithm with a complete IGM (Individual-Global-Max) function class.		 Jianhao Wang; Zhizhou Ren; Terry Liu; Yang Yu; Chongjie Zhang;
749 Learning Energy-Based Models By Diffusion Recovery Likelihood
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a diffusion recovery likelihood method to tractably learn and sample from a sequence of EBMs based on a diffusion process. We achieves high sample quality, stable long-run sampling chains and estimation of likelihood.		 Ruiqi Gao; Yang Song; Ben Poole; Ying Nian Wu; Diederik P Kingma;
750 Neural Network Extrapolations with G-invariances from A Single Environment
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This work introduces a novel learning framework for single-environment extrapolations, where invariance to transformation groups is mandatory even without evidence, unless the learner deems it inconsistent with the training data.		 S Chandra Mouli; Bruno Ribeiro;
751 Model-Based Offline Planning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This approach adapts model-based reinforcement learning to offline regimes with little data, and shows state of the art control in offline scenarios.		 Arthur Argenson; Gabriel Dulac-Arnold;
752 On Dyadic Fairness: Exploring and Mitigating Bias in Graph Connections
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A new method on the fairness of predictive relationships in graph-structured data		 Peizhao Li; Yifei Wang; Han Zhao; Pengyu Hong; Hongfu Liu;
753 Coping with Label Shift Via Distributionally Robust Optimisation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an objective to cope with label shift, and provide an adversarial algorithm to effectively optimize it.		 Jingzhao Zhang; Aditya Krishna Menon; Andreas Veit; Srinadh Bhojanapalli; Sanjiv Kumar; Suvrit Sra;
754 Faster Binary Embeddings for Preserving Euclidean Distances
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a fast binary embedding algorithm to preserve Euclidean distances among well-spread vectors and it achieves optimal bit complexity.		 Jinjie Zhang; Rayan Saab;
755 Learning and Evaluating Representations for Deep One-Class Classification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a two-stage framework for deep one-class classification, composed of state-of-the-art self-supervised representation learning followed by generative or discriminative one-class classifiers.		 Kihyuk Sohn; Chun-Liang Li; Jinsung Yoon; Minho Jin; Tomas Pfister;
756 Conditional Negative Sampling for Contrastive Learning of Visual Representations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Theoretical and experimental evidence that choosing difficult negative examples in contrastive learning can learn stronger representations as measured by several downstream tasks and image distributions.		 Mike Wu; Milan Mosse; Chengxu Zhuang; Daniel Yamins; Noah Goodman;
757 On Position Embeddings in BERT
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper amis to understand and evaluate position embeddings, especially in pretrain language models		 Benyou Wang; Lifeng Shang; Christina Lioma; Xin Jiang; Hao Yang; Qun Liu; Jakob Grue Simonsen;
758 Repurposing Pretrained Models for Robust Out-of-domain Few-Shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an alternative meta-testing procedure and combine MAML gradient steps with adversarial training and uncertainty-based stepsize adaptation.		 Namyeong Kwon; Hwidong Na; Gabriel Huang; Simon Lacoste-Julien;
759 Dataset Meta-Learning from Kernel-Ridge Regression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a meta-learning approach to distilling datasets, achieving state of the art performance for kernel-ridge regression and neural networks.		 Timothy Nguyen; Zhourong Chen; Jaehoon Lee;
760 AdaFuse: Adaptive Temporal Fusion Network for Efficient Action Recognition
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we introduce an adaptive temporal fusion network, called AdaFuse, that dynamically fuses channels from current and past feature maps for strong temporal modelling.		 Yue Meng; Rameswar Panda; Chung-Ching Lin; Prasanna Sattigeri; Leonid Karlinsky; Kate Saenko; Aude Oliva; Rogerio Feris;
761 One Network Fits All? Modular Versus Monolithic Task Formulations in Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Theoretical bounds and experimental results showing that neural networks trained with SGD can provably solve multiple, very different tasks simultaneously.		 Atish Agarwala; Abhimanyu Das; Brendan Juba; Rina Panigrahy; Vatsal Sharan; Xin Wang; Qiuyi Zhang;
762 Vector-output ReLU Neural Network Problems Are Copositive Programs: Convex Analysis of Two Layer Networks and Polynomial-time Algorithms
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We demonstrate that two-layer vector-output ReLU networks can be expressed as copositive programs, and introduce algorithms for provably finding their global optima, which are polynomial in the number of samples for a fixed data rank.		 Arda Sahiner; Tolga Ergen; John M. Pauly; Mert Pilanci;
763 In Defense of Pseudo-Labeling: An Uncertainty-Aware Pseudo-label Selection Framework for Semi-Supervised Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present an uncertainty-aware pseudo-label selection framework for semi-supervised learning which greatly reduces the noise introduced by the pseudo-labeling process.		 Mamshad Nayeem Rizve; Kevin Duarte; Yogesh S Rawat; Mubarak Shah;
764 MELR: Meta-Learning Via Modeling Episode-Level Relationships for Few-Shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This is the first work on explicitly modeling episode-level relationships for few-shot learning.		 Nanyi Fei; Zhiwu Lu; Tao Xiang; Songfang Huang;
765 Why Resampling Outperforms Reweighting for Correcting Sampling Bias with Stochastic Descents
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We explain why resampling outperforms reweighting for correcting sampling bias when stochastic gradient algorithms are used.		 Jing An; Lexing Ying; Yuhua Zhu;
766 Prediction and Generalisation Over Directed Actions By Grid Cells
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Extending existing normative prediction models of grid cells to directed transitions, and provide a unifying framework for mechanistic and normative models of grid cells.		 Changmin Yu; Timothy Behrens; Neil Burgess;
767 Hopper: Multi-hop Transformer for Spatiotemporal Reasoning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose Hopper, Multi-Hop Transformer, and CATER-h dataset to approach object-centric spatiotemporal reasoning in videos.		 Honglu Zhou; Asim Kadav; Farley Lai; Alexandru Niculescu-Mizil; Martin Renqiang Min; Mubbasir Kapadia; Hans Peter Graf;
768 Sparse Encoding for More-interpretable Feature-selecting Representations in Probabilistic Matrix Factorization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a simple modification to existing sparse matrix factorization methods to rectify widespread erroneous interpretation of the factors.		 Joshua C Chang; Patrick Fletcher; Jungmin Han; Ted L Chang; Shashaank Vattikuti; Ayah Zirikly; Bart Desmet; Carson C Chow;
769 Dance Revolution: Long-Term Dance Generation with Music Via Curriculum Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we formalize the music-driven dance generation as a sequence-to-sequence learning problem and devise a novel seq2seq architecture to efficiently process long sequences of music features and capture the fine-grained correspondence between music and dance.		 Ruozi Huang; Huang Hu; Wei Wu; Kei Sawada; Mi Zhang; Daxin Jiang;
770 PAC Confidence Predictions for Deep Neural Network Classifiers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel algorithm for constructing predicted classification confidences for DNNs that comes with provable correctness guarantees, and demonstrate how our predicted confidences can be used to enable downstream guarantees in two settings.		 Sangdon Park; Shuo Li; Osbert Bastani; Insup Lee;
771 BREEDS: Benchmarks for Subpopulation Shift
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop a methodology for constructing large-scale subpopulation shift benchmarks and use them to assess model robustness as well as the effectiveness existing robustness interventions.		 Shibani Santurkar; Dimitris Tsipras; Aleksander Madry;
772 Bypassing The Ambient Dimension: Private SGD with Gradient Subspace Identification
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we circumvent the dependence on the ambient dimension by leveraging a low-dimensional structure of gradient space in deep networks—that is, the stochastic gradients for deep nets usually stay in a low dimensional subspace in the training process.		 Yingxue Zhou; Steven Wu; Arindam Banerjee;
773 End-to-End Egospheric Spatial Memory
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: End-to-End Egospheric Spatial Memory (ESM) forms a bridge between real-time mapping and differentiable memory		 Daniel James Lenton; Stephen James; Ronald Clark; Andrew Davison;
774 Evaluating The Disentanglement of Deep Generative Models Through Manifold Topology
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Evaluate disentanglement of generative models by measuring manifold topology using persistent homology		 Sharon Zhou; Eric Zelikman; Fred Lu; Andrew Y. Ng; Gunnar E. Carlsson; Stefano Ermon;
775 SCoRe: Pre-Training for Context Representation in Conversational Semantic Parsing
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we present SCORE, a new pre-training approach for CSP tasks designed to induce representations that capture the alignment between the dialogue flow and the structural context.		 Tao Yu; Rui Zhang; Alex Polozov; Christopher Meek; Ahmed Hassan Awadallah;
776 Decoupling Global and Local Representations Via Invertible Generative Flows
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Generative Flow for Decoupled Representation Learning		 Xuezhe Ma; Xiang Kong; Shanghang Zhang; Eduard H Hovy;
777 Pre-training Text-to-Text Transformers for Concept-centric Common Sense
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose self-supervised objectives and a joint training framework to augment pre-trained language models with common sense without relying on external knowledge bases.		 Wangchunshu Zhou; Dong-Ho Lee; Ravi Kiran Selvam; Seyeon Lee; Xiang Ren;
778 Local Search Algorithms for Rank-Constrained Convex Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Efficient greedy and local search algorithms for optimizing a convex objective under a rank constraint.		 Kyriakos Axiotis; Maxim Sviridenko;
779 Combining Label Propagation and Simple Models Out-performs Graph Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Here, we show that for many standard transductive node classification benchmarks, we can ex-ceed or match the performance of state-of-the-art GNNs by combining shallow multilayer perceptrons models (that ignore the graph structure entirely) with two simple postprocessings for correlation in the label structure: (i) an ?error correlation? that spreads residual errors in training data to correct errors in test data and(ii) an ?prediction correlation? that smooths the predictions on the test data.		 Qian Huang; Horace He; Abhay Singh; Ser-Nam Lim; Austin Benson;
780 Supervised Contrastive Learning for Pre-trained Language Model Fine-tuning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Driven by the intuition that good generalization requires capturing the similarity between examples in one class and contrasting them with examples in other classes, we propose a supervised contrastive learning (SCL) objective for the fine-tuning stage.		 Beliz Gunel; Jingfei Du; Alexis Conneau; Veselin Stoyanov;
781 SAFENet: A Secure, Accurate and Fast Neural Network Inference
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose SAFENet that supports automatic channel-wise activation approximation to enable a Secure, Accurate and Fast nEural Network inference service.		 Qian Lou; Yilin Shen; Hongxia Jin; Lei Jiang;
782 Provably Robust Classification of Adversarial Examples with Detection
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a joint classifier/detector training scheme with provable performance guarantees against adversarial perturbations.		 Fatemeh Sheikholeslami; Ali Lotfi; J Zico Kolter;
783 Saliency Is A Possible Red Herring When Diagnosing Poor Generalization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We controlled feature construction on images using masks to help models generalize to test distributions with covariate shift and noticed that it didn’t affect the saliency maps in the way one would expect even though it improved generalization.		 Joseph D Viviano; Becks Simpson; Francis Dutil; Yoshua Bengio; Joseph Paul Cohen;
784 Fourier Neural Operator for Parametric Partial Differential Equations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel neural operator based on Fourier transformation for learning partial differential equations.		 Zongyi Li; Nikola Borislavov Kovachki; Kamyar Azizzadenesheli; Burigede liu; Kaushik Bhattacharya; Andrew Stuart; Anima Anandkumar;
785 Combining Ensembles and Data Augmentation Can Harm Your Calibration
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We found that combining ensembles and data augmentation worsens calibration than applying them individually, and we proposed a simple fix to it.		 Yeming Wen; Ghassen Jerfel; Rafael Muller; Michael W Dusenberry; Jasper Snoek; Balaji Lakshminarayanan; Dustin Tran;
786 SOLAR: Sparse Orthogonal Learned and Random Embeddings
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a distributed training scheme to learn high dimensional sparse embeddings that are much better than dense embeddings on both precision and speed.		 Tharun Medini; Beidi Chen; Anshumali Shrivastava;
787 Efficient Empowerment Estimation for Unsupervised Stabilization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an efficient estimation of empowerment which is demonstrated on unsupervised stabilization of dynamical systems, and compared to the existing relevant methods.		 Ruihan Zhao; Kevin Lu; Pieter Abbeel; Stas Tiomkin;
788 More Or Less: When and How to Build Convolutional Neural Network Ensembles
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that when we perform a holistic assessment, we uncover a wide design space, where ensembles not only provide better accuracy but also train and deploy with fewer resources than comparable single convolutional network models.		 Abdul Wasay; Stratos Idreos;
789 VTNet: Visual Transformer Network for Object Goal Navigation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we introduce a Visual Transformer Network (VTNet) for learning informative visual representation in navigation.		 Heming Du; Xin Yu; Liang Zheng;
790 Class Normalization for Zero-Shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We develop theoretical understanding of signal normalization inside zero-shot learning models, propose a novel normalization scheme and use it to achieve SotA ZSL performance with a simple MLP		 Ivan Skorokhodov; Mohamed Elhoseiny;
791 Batch Reinforcement Learning Through Continuation Method
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this work, we propose a simple yet effective policy iteration approach to batch RL using global optimization techniques known as continuation.		 Yijie Guo; Shengyu Feng; Nicolas Le Roux; Ed Chi; Honglak Lee; Minmin Chen;
792 Towards Resolving The Implicit Bias of Gradient Descent for Matrix Factorization: Greedy Low-Rank Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove that for depth-2 matrix factorization, gradient flow with infinitesimal initialization is mathematically equivalent to a simple heuristic rank minimization algorithm, Greedy Low-Rank Learning, under some reasonable assumptions.		 Zhiyuan Li; Yuping Luo; Kaifeng Lyu;
793 Training BatchNorm and Only BatchNorm: On The Expressive Power of Random Features in CNNs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the role and expressive power of learned affine parameters that transform features by freezing all weights at their random initializations and training only BatchNorm.		 Jonathan Frankle; David J. Schwab; Ari S. Morcos;
794 CompOFA � Compound Once-For-All Networks for Faster Multi-Platform Deployment
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: CNN design-space and system insights for faster latency-guided training and searching of models for diverse deployment targets.		 Manas Sahni; Shreya Varshini; Alind Khare; Alexey Tumanov;
795 Adaptive and Generative Zero-Shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We address the problem of generalized zero-shot learning (GZSL) where the task is to predict the class label of a target image whether its label belongs to the seen or unseen category.		 Yu-Ying Chou; Hsuan-Tien Lin; Tyng-Luh Liu;
796 Learning to Make Decisions Via Submodular Regularization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A data-driven sequential decision making framework based on a novel submodular-regularized loss.		 Ayya Alieva; Aiden Aceves; Jialin Song; Stephen Mayo; Yisong Yue; Yuxin Chen;
797 Generating Furry Cars: Disentangling Object Shape and Appearance Across Multiple Domains
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present a framework for multi-domain disentanglement, facilitating transfer of appearance from one domain to another.		 Utkarsh Ojha; Krishna Kumar Singh; Yong Jae Lee;
798 Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Studies feature matrix rank collapse (i.e. implicit regularization) in deep Q-learning methods.		 Aviral Kumar; Rishabh Agarwal; Dibya Ghosh; Sergey Levine;
799 Disentangling 3D Prototypical Networks for Few-Shot Concept Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present neural architectures that disentangle RGB-D images into objects? shapes and styles and a map of the background scene, and explore their applications for few-shot 3D object detection and few-shot concept classification.		 Mihir Prabhudesai; Shamit Lal; Darshan Patil; Hsiao-Yu Tung; Adam W Harley; Katerina Fragkiadaki;
800 Anytime Sampling for Autoregressive Models Via Ordered Autoencoding
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new family of autoregressive model that enables anytime sampling		 Yilun Xu; Yang Song; Sahaj Garg; Linyuan Gong; Rui Shu; Aditya Grover; Stefano Ermon;
801 HyperDynamics: Generating Expert Dynamics Models By Observation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose HyperDynamics, a framework that conditions on an agent?s interactions with the environment and optionally its visual observations, and generates the parameters of neural dynamics models based on inferred properties of the dynamical system.		 Zhou Xian; Shamit Lal; Hsiao-Yu Tung; Emmanouil Antonios Platanios; Katerina Fragkiadaki;
802 Improving Zero-Shot Voice Style Transfer Via Disentangled Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An information-theoretic disentangled representation learning framework for zero-shot voice style transfer.		 Siyang Yuan; Pengyu Cheng; Ruiyi Zhang; Weituo Hao; Zhe Gan; Lawrence Carin;
803 GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Language model pre-training for table semantic parsing.		 Tao Yu; Chien-Sheng Wu; Xi Victoria Lin; bailin wang; Yi Chern Tan; Xinyi Yang; Dragomir Radev; richard socher; Caiming Xiong;
804 Estimating Lipschitz Constants of Monotone Deep Equilibrium Models
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Monotone deep equilibrium models have Lipschitz constants which are simple to bound and small relative to those of standard DNNs, which suffer with depth.		 Chirag Pabbaraju; Ezra Winston; J Zico Kolter;
805 Estimating Informativeness of Samples with Smooth Unique Information
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We define, both in weight-space and function-space, a notion of unique information that an individual sample provides to the training of a deep network and show how to compute it efficiently large networks using a linearization of the model.		 Hrayr Harutyunyan; Alessandro Achille; Giovanni Paolini; Orchid Majumder; Avinash Ravichandran; Rahul Bhotika; Stefano Soatto;
806 NBDT: Neural-Backed Decision Tree
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Neural-Backed Decision Trees improve interpretability and accuracy: (1) out-generalize, improve, and match or outperform baseline neural networks; (2) show visual evidence of generalization, reveal ambiguous ImageNet labels, and improve human trust.		 Alvin Wan; Lisa Dunlap; Daniel Ho; Jihan Yin; Scott Lee; Suzanne Petryk; Sarah Adel Bargal; Joseph E. Gonzalez;
807 Accurate Learning of Graph Representations with Graph Multiset Pooling
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel graph pooling method for graph representation learning, that considers multiset with attention-based operations.		 Jinheon Baek; Minki Kang; Sung Ju Hwang;
808 Byzantine-Resilient Non-Convex Stochastic Gradient Descent
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We give a new algorithm for Byzantine-resilient non-convex distributed optimization, with strong theoretical guarantees, which improves on the performance of prior methods for training deep neural networks against Byzantine attacks.		 Dan Alistarh; Zeyuan Allen-Zhu; Faeze Ebrahimianghazani; Jerry Li;
809 MetaNorm: Learning to Normalize Few-Shot Batches Across Domains
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose MetaNorm, a simple yet effective meta-learning normalization approach that learns adaptive statistics for few-shot classification and domain generalization		 Yingjun Du; Xiantong Zhen; Ling Shao; Cees G. M. Snoek;
810 Large Batch Simulation for Deep Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: The key idea of our approach is to design a 3D renderer and environment simulator around the principle of “batch simulation”: accepting and executing large batches of requests simultaneously.		 Brennan Shacklett; Erik Wijmans; Aleksei Petrenko; Manolis Savva; Dhruv Batra; Vladlen Koltun; Kayvon Fatahalian;
811 Personalized Federated Learning with First Order Model Optimization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new federated learning framework that efficiently computes a personalized weighted combination of available models for each client, outperforming existing work for personalized federated learning.		 Michael Zhang; Karan Sapra; Sanja Fidler; Serena Yeung; Jose M. Alvarez;
812 Combining Physics and Machine Learning for Network Flow Estimation
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Paper solves missing flow estimation problem using bilevel optimization and deep learning.		 Arlei Lopes da Silva; Furkan Kocayusufoglu; Saber Jafarpour; Francesco Bullo; Ananthram Swami; Ambuj Singh;
813 Knowledge Distillation As Semiparametric Inference
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Viewing knowledge distillation as a semiparametric inference problem leads to improved generalization guarantees of the distillation process		 Tri Dao; Govinda M Kamath; Vasilis Syrgkanis; Lester Mackey;
814 Learning Value Functions in Deep Policy Gradients Using Residual Variance
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a method to improve the learning of the critic in the actor-critic framework.		 Yannis Flet-Berliac; reda ouhamma; odalric-ambrym maillard; Philippe Preux;
815 Randomized Ensembled Double Q-Learning: Learning Fast Without A Model
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose and analyze a novel model-free algorithm that achieves strong performance with a high update-to-data ratio.		 Xinyue Chen; Che Wang; Zijian Zhou; Keith W. Ross;
816 DECENTRALIZED ATTRIBUTION OF GENERATIVE MODELS
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper investigates the feasibility of decentralized attribution of generative models.		 Changhoon Kim; Yi Ren; Yezhou Yang;
817 Constellation Nets for Few-Shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We tackle the few-shot learning problem by introducing an explicit cell feature clustering procedure with relation learning via self-attention.		 Weijian Xu; yifan xu; Huaijin Wang; Zhuowen Tu;
818 Adapting to Reward Progressivity Via Spectral Reinforcement Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we identify a problem with value-based deep reinforcement learning that has not previously been investigated — namely, reward progressivity — and propose an approach that addresses it via magnitudinal decomposition of the reward.		 Michael Dann; John Thangarajah;
819 TropEx: An Algorithm for Extracting Linear Terms in Deep Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an algorithm for extracting linear terms of piecewise linear deep neural network functions and apply it to study differences between convolutional and fully-connected networks.		 Martin Trimmel; Henning Petzka; Cristian Sminchisescu;
820 Reset-Free Lifelong Learning with Skill-Space Planning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose \textit{Lifelong Skill Planning} (LiSP), an algorithmic framework for lifelong RL based on planning in an abstract space of higher-order skills.		 Kevin Lu; Aditya Grover; Pieter Abbeel; Igor Mordatch;
821 Robust Learning of Fixed-Structure Bayesian Networks in Nearly-Linear Time
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We give the first nearly-linear time algorithm for the robust learning of fixed-structure Bayesian networks.		 Yu Cheng; Honghao Lin;
822 Teaching Temporal Logics to Neural Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study two fundamental questions in neuro-symbolic computing: can deep learning tackle challenging problems in logics end-to-end, and can neural networks learn the semantics of logics.		 Christopher Hahn; Frederik Schmitt; Jens U. Kreber; Markus Norman Rabe; Bernd Finkbeiner;
823 Spatially Structured Recurrent Modules
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We model a dynamical system as a collection of recurrent modules that interact according to a spatially informed but learned topology.		 Nasim Rahaman; Anirudh Goyal; Muhammad Waleed Gondal; Manuel Wuthrich; Stefan Bauer; Yash Sharma; Yoshua Bengio; Bernhard Sch�lkopf;
824 Bayesian Few-Shot Classification with One-vs-Each P�lya-Gamma Augmented Gaussian Processes
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a Gaussian process approach to few-shot classification based on the one-vs-each softmax approximation and P?lya-gamma augmentation, and demonstrate competitive few-shot accuracy and strong uncertainty quantification.		 Jake Snell; Richard Zemel;
825 Parameter-based Value Functions
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose value functions whose inputs include the policy parameters and which can generalize across different policies		 Francesco Faccio; Louis Kirsch; J�rgen Schmidhuber;
826 Hyperbolic Neural Networks++
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We present novel methods for constructing hyperbolic neural network architectures in the Poincar? ball model, including a parameter-reduced MLR, geodesic-aware FC layers, convolutional layers, and attention mechanisms.		 Ryohei Shimizu; YUSUKE Mukuta; Tatsuya Harada;
827 Neural Jump Ordinary Differential Equations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: On-line prediction of irregularly-observed time series using neural ODE with jumps with theoretical and empirical justification.		 Calypso Herrera; Florian Krach; Josef Teichmann;
828 Seq2Tens: An Efficient Representation of Sequences By Low-Rank Tensor Projections
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An Efficient Representation of Sequences by Low-Rank Tensor Projections		 Csaba Toth; Patric Bonnier; Harald Oberhauser;
829 Efficient Fully-Offline Meta-Reinforcement Learning Via Distance Metric Learning and Behavior Regularization
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A model-free, end-to-end fully-offline meta-RL algorithm designed to maximize practicality, performance and sample/computational efficiency.		 Lanqing Li; Rui Yang; Dijun Luo;
830 On The Curse of Memory in Recurrent Neural Networks: Approximation and Optimization Analysis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We study the approximation properties and optimization dynamics of RNNs in the linear setting, where we uncover precisely the adverse effect of memory on learning.		 Zhong Li; Jiequn Han; Weinan E; Qianxiao Li;
831 Generating Adversarial Computer Programs Using Optimized Obfuscations
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A differentiable generator of adversarial computer programs which can deceive ML models trained on computer programs.		 Shashank Srikant; Sijia Liu; Tamara Mitrovska; Shiyu Chang; Quanfu Fan; Gaoyuan Zhang; Una-May O’Reilly;
832 BOIL: Towards Representation Change for Few-shot Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel meta-learning algorithm, BOIL, based on representation change.		 Jaehoon Oh; Hyungjun Yoo; ChangHwan Kim; Se-Young Yun;
833 Towards Understanding and Improving Dropout in Game Theory
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We prove and improve the utility of the dropout operation from a game-theoretic view.		 Hao Zhang; Sen Li; YinChao Ma; Mingjie Li; Yichen Xie; Quanshi Zhang;
834 Representation Learning Via Invariant Causal Mechanisms
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a new self-supervised objective with an explicit invariance regularizer and provide an alternative explanation for the success of contrastive learning using causality; we outperform competing methods on ImageNet and Atari.		 Jovana Mitrovic; Brian McWilliams; Jacob C Walker; Lars Holger Buesing; Charles Blundell;
835 Fooling A Complete Neural Network Verifier
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose an attack (along with a defense) to fool complete verification based on exploiting numerical errors.		 D�niel Zombori; Bal�zs B�nhelyi; Tibor Csendes; Istv�n Megyeri; M�rk Jelasity;
836 CPR: Classifier-Projection Regularization for Continual Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We devise wide local minima promoting regularization term for continual learning.		 Sungmin Cha; Hsiang Hsu; Taebaek Hwang; Flavio Calmon; Taesup Moon;
837 CO2: Consistent Contrast for Unsupervised Visual Representation Learning
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We introduce a consistency regularization term into the current contrastive learning framework.		 Chen Wei; Huiyu Wang; Wei Shen; Alan Yuille;
838 GAN2GAN: Generative Noise Learning for Blind Denoising with Single Noisy Images
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We devise GAN2GAN method that trains a blind denoiser solely based on the single noisy images.		 Sungmin Cha; Taeeon Park; Byeongjoon Kim; Jongduk Baek; Taesup Moon;
839 Learning Subgoal Representations with Slow Dynamics
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a slowness objective to learn subgoal representations in hierarchical reinforcement learning.		 Siyuan Li; Lulu Zheng; Jianhao Wang; Chongjie Zhang;
840 Bowtie Networks: Generative Modeling for Joint Few-Shot Recognition and Novel-View Synthesis
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a novel feedback-based bowtie network to learn a shared generative model for joint few-shot recognition and novel-view synthesis, consistently and significantly improving performance for both tasks, especially in the low-data regime.		 Zhipeng Bao; Yu-Xiong Wang; Martial Hebert;
841 Taming GANs with Lookahead-Minmax
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A novel optimizer for GANs and games in general.		 Tatjana Chavdarova; Matteo Pagliardini; Sebastian U Stich; Fran�ois Fleuret; Martin Jaggi;
842 Certify Or Predict: Boosting Certified Robustness with Compositional Architectures
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a compositional network architecture boosting the certified robustness of an accurate state-of-the-art network, by combining it with a shallow, provable network using a certified, adaptive selection mechanism.		 Mark Niklas Mueller; Mislav Balunovic; Martin Vechev;
843 New Bounds For Distributed Mean Estimation and Variance Reduction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We provide optimal algorithms and lower bounds for distributed mean estimation and variance reduction via a new connection to lattice theory, and show that this technique can be used to improve upon current approaches in practice.		 Peter Davies; Vijaykrishna Gurunanthan; Niusha Moshrefi; Saleh Ashkboos; Dan Alistarh;
844 Interpretable Neural Architecture Search Via Bayesian Optimisation with Weisfeiler-Lehman Kernels
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a NAS method that is sample-efficient, highly performant and interpretable.		 Xingchen Wan; Binxin Ru; Xiaowen Dong; Michael Osborne;
845 A Discriminative Gaussian Mixture Model with Sparsity
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: A sparse classifier based on a discriminative Gaussian mixture model, which can also be embedded into a neural network.		 Hideaki Hayashi; Seiichi Uchida;
846 Communication in Multi-Agent Reinforcement Learning: Intention Sharing
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: This paper propose a new communication scheme named intention sharing to enhance the coordination among agents.		 Woojun Kim; Jongeui Park; Youngchul Sung;
847 Is Attention Better Than Matrix Decomposition?
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Our intriguing finding is that self-attention is not better than the matrix decomposition (MD) model developed 20 years ago regarding the performance and computational cost for encoding the long-distance dependencies.		 Zhengyang Geng; Meng-Hao Guo; Hongxu Chen; Xia Li; Ke Wei; Zhouchen Lin;
848 Fast and Complete: Enabling Complete Neural Network Verification with Rapid and Massively Parallel Incomplete Verifiers
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We use fast bound propagation methods on GPUs for complete neural network verification and achieve large speedup compared to SOTA.		 Kaidi Xu; Huan Zhang; Shiqi Wang; Yihan Wang; Suman Jana; Xue Lin; Cho-Jui Hsieh;
849 A Geometric Analysis of Deep Generative Image Models and Its Applications
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We developed tools to compute the metric tensor of image manifold learnt by GANs, empirically analyzed their geometry, and found this knowledge useful to GAN inversion and finding interpretable axes.		 Binxu Wang; Carlos R Ponce;
850 Solving Compositional Reinforcement Learning Problems Via Task Reduction
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose a deep RL algorithm for learning compositional strategies to solve sparse-reward continuous-control problems.		 Yunfei Li; Huazhe Xu; Yilin Wu; Xiaolong Wang; Yi Wu;
851 ARMOURED: Adversarially Robust MOdels Using Unlabeled Data By REgularizing Diversity
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: ARMOURED is a novel technique for adversarially robust learning that elegantly unifies semi-supervised learning and diversity regularization through a multi-view learning framework.		 Kangkang Lu; Cuong Manh Nguyen; Xun Xu; Kiran Chari; Yu Jing Goh; Chuan-Sheng Foo;
852 Acting in Delayed Environments with Non-Stationary Markov Policies
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we derive theoretical results on execution-delay MDPs, and devise a DQN-based algorithm to empirically tackle this setup.		 Gal Dalal; Esther Derman; Shie Mannor;
853 Overfitting for Fun and Profit: Instance-Adaptive Data Compression
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We show that we can finetune an entire data compression model on a single instance, and improve the rate-distortion performance, taking into account the additional costs for sending the model updates.		 Ties van Rozendaal; Iris AM Huijben; Taco Cohen;
854 LEARNABLE EMBEDDING SIZES FOR RECOMMENDER SYSTEMS
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Learning flexible feature-aware embedding sizes effectively and efficiently for recommendation models.		 Siyi Liu; Chen Gao; Yihong Chen; Depeng Jin; Yong Li;
855 Generative Scene Graph Networks
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We propose the first object-centric generative model capable of learning part-object relationships from multi-object scenes without access to pre-segmented parts.		 Fei Deng; Zhuo Zhi; Donghun Lee; Sungjin Ahn;
856 Deconstructing The Regularization of BatchNorm
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: We deconstruct the regularization effect of batch normalization and show that preventing explosive growth at the final layer at initialization and during training can explain most of BatchNorm’s generalization boost.		 Yann Dauphin; Ekin Dogus Cubuk;
857 PolarNet: Learning to Optimize Polar Keypoints for Keypoint Based Object Detection
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: In this paper, we propose a new anchor-free keypoint based detector “PolarNet, where keypoints are represented as a set of Polar coordinates instead of Cartesian coordinates.		 Wu Xiongwei; Steven HOI; Doyen Sahoo;
858 Simple Spectral Graph Convolution
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: It is a simple and efficient method for graph convolution based on the Markov Diffusion Kernel which works well in different tasks under unsupervised, semi-supervised and supervised setting.		 Hao Zhu; Piotr Koniusz;
859 Explainable Subgraph Reasoning for Forecasting on Temporal Knowledge Graphs
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: Especially, we propose a temporal relational attention mechanism and a novel reverse representation update scheme to guide the extraction of an enclosing subgraph around the query.		 Zhen Han; Peng Chen; Yunpu Ma; Volker Tresp;
860 EVALUATION OF NEURAL ARCHITECTURES TRAINED WITH SQUARE LOSS VS CROSS-ENTROPY IN CLASSIFICATION TASKS
Related Papers   Related Patents   Related Grants   Related Orgs   Related Experts   Details
Highlight: An experimental evaluation of neural architectures in classification tasks shows that training with square loss produces better results than the cross-entropy in the majority of NLP and ASR experiments.		 Like Hui; Mikhail Belkin;