paper_id
stringlengths
9
24
title
stringlengths
10
192
paper_url
stringlengths
44
59
authors
listlengths
1
23
type
stringclasses
0 values
primary_area
stringclasses
0 values
abstract
large_stringlengths
82
4.49k
keywords
listlengths
0
0
TL;DR
large_stringclasses
0 values
submission_number
int64
1
1.18k
arxiv_id
stringlengths
10
10
arxiv_id_source
stringclasses
2 values
v139/raileanu21a
Decoupling Value and Policy for Generalization in Reinforcement Learning
https://proceedings.mlr.press/v139/raileanu21a.html
[ "Roberta Raileanu", "Rob Fergus" ]
null
null
Standard deep reinforcement learning algorithms use a shared representation for the policy and value function, especially when training directly from images. However, we argue that more information is needed to accurately estimate the value function than to learn the optimal policy. Consequently, the use of a shared re...
[]
null
801
2102.10330
title_snapshot
v139/rajagopalan21a
Hierarchical Clustering of Data Streams: Scalable Algorithms and Approximation Guarantees
https://proceedings.mlr.press/v139/rajagopalan21a.html
[ "Anand Rajagopalan", "Fabio Vitale", "Danny Vainstein", "Gui Citovsky", "Cecilia M Procopiuc", "Claudio Gentile" ]
null
null
We investigate the problem of hierarchically clustering data streams containing metric data in R^d. We introduce a desirable invariance property for such algorithms, describe a general family of hyperplane-based methods enjoying this property, and analyze two scalable instances of this general family against recently p...
[]
null
802
null
null
v139/rakotomamonjy21a
Differentially Private Sliced Wasserstein Distance
https://proceedings.mlr.press/v139/rakotomamonjy21a.html
[ "Alain Rakotomamonjy", "Ralaivola Liva" ]
null
null
Developing machine learning methods that are privacy preserving is today a central topic of research, with huge practical impacts. Among the numerous ways to address privacy-preserving learning, we here take the perspective of computing the divergences between distributions under the Differential Privacy (DP) framework...
[]
null
803
2107.01848
title_snapshot
v139/ramesh21a
Zero-Shot Text-to-Image Generation
https://proceedings.mlr.press/v139/ramesh21a.html
[ "Aditya Ramesh", "Mikhail Pavlov", "Gabriel Goh", "Scott Gray", "Chelsea Voss", "Alec Radford", "Mark Chen", "Ilya Sutskever" ]
null
null
Text-to-image generation has traditionally focused on finding better modeling assumptions for training on a fixed dataset. These assumptions might involve complex architectures, auxiliary losses, or side information such as object part labels or segmentation masks supplied during training. We describe a simple approach...
[]
null
804
2102.12092
title_snapshot
v139/rangapuram21a
End-to-End Learning of Coherent Probabilistic Forecasts for Hierarchical Time Series
https://proceedings.mlr.press/v139/rangapuram21a.html
[ "Syama Sundar Rangapuram", "Lucien D Werner", "Konstantinos Benidis", "Pedro Mercado", "Jan Gasthaus", "Tim Januschowski" ]
null
null
This paper presents a novel approach for hierarchical time series forecasting that produces coherent, probabilistic forecasts without requiring any explicit post-processing reconciliation. Unlike the state-of-the-art, the proposed method simultaneously learns from all time series in the hierarchy and incorporates the r...
[]
null
805
null
null
v139/rao21a
MSA Transformer
https://proceedings.mlr.press/v139/rao21a.html
[ "Roshan M Rao", "Jason Liu", "Robert Verkuil", "Joshua Meier", "John Canny", "Pieter Abbeel", "Tom Sercu", "Alexander Rives" ]
null
null
Unsupervised protein language models trained across millions of diverse sequences learn structure and function of proteins. Protein language models studied to date have been trained to perform inference from individual sequences. The longstanding approach in computational biology has been to make inferences from a fami...
[]
null
806
null
null
v139/rasul21a
Autoregressive Denoising Diffusion Models for Multivariate Probabilistic Time Series Forecasting
https://proceedings.mlr.press/v139/rasul21a.html
[ "Kashif Rasul", "Calvin Seward", "Ingmar Schuster", "Roland Vollgraf" ]
null
null
In this work, we propose TimeGrad, an autoregressive model for multivariate probabilistic time series forecasting which samples from the data distribution at each time step by estimating its gradient. To this end, we use diffusion probabilistic models, a class of latent variable models closely connected to score matchi...
[]
null
807
2101.12072
title_snapshot
v139/ratzlaff21a
Generative Particle Variational Inference via Estimation of Functional Gradients
https://proceedings.mlr.press/v139/ratzlaff21a.html
[ "Neale Ratzlaff", "Qinxun Bai", "Li Fuxin", "Wei Xu" ]
null
null
Recently, particle-based variational inference (ParVI) methods have gained interest because they can avoid arbitrary parametric assumptions that are common in variational inference. However, many ParVI approaches do not allow arbitrary sampling from the posterior, and the few that do allow such sampling suffer from sub...
[]
null
808
2103.01291
title_snapshot
v139/raviv21a
Enhancing Robustness of Neural Networks through Fourier Stabilization
https://proceedings.mlr.press/v139/raviv21a.html
[ "Netanel Raviv", "Aidan Kelley", "Minzhe Guo", "Yevgeniy Vorobeychik" ]
null
null
Despite the considerable success of neural networks in security settings such as malware detection, such models have proved vulnerable to evasion attacks, in which attackers make slight changes to inputs (e.g., malware) to bypass detection. We propose a novel approach, Fourier stabilization, for designing evasion-robus...
[]
null
809
2106.04435
title_snapshot
v139/rawat21a
Disentangling Sampling and Labeling Bias for Learning in Large-output Spaces
https://proceedings.mlr.press/v139/rawat21a.html
[ "Ankit Singh Rawat", "Aditya K Menon", "Wittawat Jitkrittum", "Sadeep Jayasumana", "Felix Yu", "Sashank Reddi", "Sanjiv Kumar" ]
null
null
Negative sampling schemes enable efficient training given a large number of classes, by offering a means to approximate a computationally expensive loss function that takes all labels into account. In this paper, we present a new connection between these schemes and loss modification techniques for countering label imb...
[]
null
810
2105.05736
title_snapshot
v139/raychaudhuri21a
Cross-domain Imitation from Observations
https://proceedings.mlr.press/v139/raychaudhuri21a.html
[ "Dripta S. Raychaudhuri", "Sujoy Paul", "Jeroen Vanbaar", "Amit K. Roy-Chowdhury" ]
null
null
Imitation learning seeks to circumvent the difficulty in designing proper reward functions for training agents by utilizing expert behavior. With environments modeled as Markov Decision Processes (MDP), most of the existing imitation algorithms are contingent on the availability of expert demonstrations in the same MDP...
[]
null
811
2105.10037
title_snapshot
v139/razin21a
Implicit Regularization in Tensor Factorization
https://proceedings.mlr.press/v139/razin21a.html
[ "Noam Razin", "Asaf Maman", "Nadav Cohen" ]
null
null
Recent efforts to unravel the mystery of implicit regularization in deep learning have led to a theoretical focus on matrix factorization — matrix completion via linear neural network. As a step further towards practical deep learning, we provide the first theoretical analysis of implicit regularization in tensor facto...
[]
null
812
2102.09972
title_snapshot
v139/refinetti21a
Align, then memorise: the dynamics of learning with feedback alignment
https://proceedings.mlr.press/v139/refinetti21a.html
[ "Maria Refinetti", "Stéphane D’Ascoli", "Ruben Ohana", "Sebastian Goldt" ]
null
null
Direct Feedback Alignment (DFA) is emerging as an efficient and biologically plausible alternative to backpropagation for training deep neural networks. Despite relying on random feedback weights for the backward pass, DFA successfully trains state-of-the-art models such as Transformers. On the other hand, it notorious...
[]
null
813
2011.12428
title_snapshot
v139/refinetti21b
Classifying high-dimensional Gaussian mixtures: Where kernel methods fail and neural networks succeed
https://proceedings.mlr.press/v139/refinetti21b.html
[ "Maria Refinetti", "Sebastian Goldt", "Florent Krzakala", "Lenka Zdeborova" ]
null
null
A recent series of theoretical works showed that the dynamics of neural networks with a certain initialisation are well-captured by kernel methods. Concurrent empirical work demonstrated that kernel methods can come close to the performance of neural networks on some image classification tasks. These results raise the ...
[]
null
814
2102.11742
title_snapshot
v139/rematas21a
Sharf: Shape-conditioned Radiance Fields from a Single View
https://proceedings.mlr.press/v139/rematas21a.html
[ "Konstantinos Rematas", "Ricardo Martin-Brualla", "Vittorio Ferrari" ]
null
null
We present a method for estimating neural scenes representations of objects given only a single image. The core of our method is the estimation of a geometric scaffold for the object and its use as a guide for the reconstruction of the underlying radiance field. Our formulation is based on a generative process that fir...
[]
null
815
2102.08860
title_snapshot
v139/ren21a
LEGO: Latent Execution-Guided Reasoning for Multi-Hop Question Answering on Knowledge Graphs
https://proceedings.mlr.press/v139/ren21a.html
[ "Hongyu Ren", "Hanjun Dai", "Bo Dai", "Xinyun Chen", "Michihiro Yasunaga", "Haitian Sun", "Dale Schuurmans", "Jure Leskovec", "Denny Zhou" ]
null
null
Answering complex natural language questions on knowledge graphs (KGQA) is a challenging task. It requires reasoning with the input natural language questions as well as a massive, incomplete heterogeneous KG. Prior methods obtain an abstract structured query graph/tree from the input question and traverse the KG for a...
[]
null
816
null
null
v139/ren21b
Interpreting and Disentangling Feature Components of Various Complexity from DNNs
https://proceedings.mlr.press/v139/ren21b.html
[ "Jie Ren", "Mingjie Li", "Zexu Liu", "Quanshi Zhang" ]
null
null
This paper aims to define, visualize, and analyze the feature complexity that is learned by a DNN. We propose a generic definition for the feature complexity. Given the feature of a certain layer in the DNN, our method decomposes and visualizes feature components of different complexity orders from the feature. The fea...
[]
null
817
2006.15920
title_snapshot
v139/ren21c
Integrated Defense for Resilient Graph Matching
https://proceedings.mlr.press/v139/ren21c.html
[ "Jiaxiang Ren", "Zijie Zhang", "Jiayin Jin", "Xin Zhao", "Sixing Wu", "Yang Zhou", "Yelong Shen", "Tianshi Che", "Ruoming Jin", "Dejing Dou" ]
null
null
A recent study has shown that graph matching models are vulnerable to adversarial manipulation of their input which is intended to cause a mismatching. Nevertheless, there is still a lack of a comprehensive solution for further enhancing the robustness of graph matching against adversarial attacks. In this paper, we id...
[]
null
818
null
null
v139/richter21a
Solving high-dimensional parabolic PDEs using the tensor train format
https://proceedings.mlr.press/v139/richter21a.html
[ "Lorenz Richter", "Leon Sallandt", "Nikolas Nüsken" ]
null
null
High-dimensional partial differential equations (PDEs) are ubiquitous in economics, science and engineering. However, their numerical treatment poses formidable challenges since traditional grid-based methods tend to be frustrated by the curse of dimensionality. In this paper, we argue that tensor trains provide an app...
[]
null
819
2102.11830
title_snapshot
v139/rizk21a
Best Arm Identification in Graphical Bilinear Bandits
https://proceedings.mlr.press/v139/rizk21a.html
[ "Geovani Rizk", "Albert Thomas", "Igor Colin", "Rida Laraki", "Yann Chevaleyre" ]
null
null
We introduce a new graphical bilinear bandit problem where a learner (or a \emph{central entity}) allocates arms to the nodes of a graph and observes for each edge a noisy bilinear reward representing the interaction between the two end nodes. We study the best arm identification problem in which the learner wants to f...
[]
null
820
2012.07641
title_snapshot
v139/roddenberry21a
Principled Simplicial Neural Networks for Trajectory Prediction
https://proceedings.mlr.press/v139/roddenberry21a.html
[ "T. Mitchell Roddenberry", "Nicholas Glaze", "Santiago Segarra" ]
null
null
We consider the construction of neural network architectures for data on simplicial complexes. In studying maps on the chain complex of a simplicial complex, we define three desirable properties of a simplicial neural network architecture: namely, permutation equivariance, orientation equivariance, and simplicial aware...
[]
null
821
2102.10058
title_snapshot
v139/roeder21a
On Linear Identifiability of Learned Representations
https://proceedings.mlr.press/v139/roeder21a.html
[ "Geoffrey Roeder", "Luke Metz", "Durk Kingma" ]
null
null
Identifiability is a desirable property of a statistical model: it implies that the true model parameters may be estimated to any desired precision, given sufficient computational resources and data. We study identifiability in the context of representation learning: discovering nonlinear data representations that are ...
[]
null
822
2007.00810
title_snapshot
v139/rolf21a
Representation Matters: Assessing the Importance of Subgroup Allocations in Training Data
https://proceedings.mlr.press/v139/rolf21a.html
[ "Esther Rolf", "Theodora T Worledge", "Benjamin Recht", "Michael Jordan" ]
null
null
Collecting more diverse and representative training data is often touted as a remedy for the disparate performance of machine learning predictors across subpopulations. However, a precise framework for understanding how dataset properties like diversity affect learning outcomes is largely lacking. By casting data colle...
[]
null
823
2103.03399
title_snapshot
v139/romac21a
TeachMyAgent: a Benchmark for Automatic Curriculum Learning in Deep RL
https://proceedings.mlr.press/v139/romac21a.html
[ "Clément Romac", "Rémy Portelas", "Katja Hofmann", "Pierre-Yves Oudeyer" ]
null
null
Training autonomous agents able to generalize to multiple tasks is a key target of Deep Reinforcement Learning (DRL) research. In parallel to improving DRL algorithms themselves, Automatic Curriculum Learning (ACL) study how teacher algorithms can train DRL agents more efficiently by adapting task selection to their ev...
[]
null
824
2103.09815
title_snapshot
v139/rosca21a
Discretization Drift in Two-Player Games
https://proceedings.mlr.press/v139/rosca21a.html
[ "Mihaela C Rosca", "Yan Wu", "Benoit Dherin", "David Barrett" ]
null
null
Gradient-based methods for two-player games produce rich dynamics that can solve challenging problems, yet can be difficult to stabilize and understand. Part of this complexity originates from the discrete update steps given by simultaneous or alternating gradient descent, which causes each player to drift away from th...
[]
null
825
2105.13922
title_snapshot
v139/rosenfeld21a
On the Predictability of Pruning Across Scales
https://proceedings.mlr.press/v139/rosenfeld21a.html
[ "Jonathan S Rosenfeld", "Jonathan Frankle", "Michael Carbin", "Nir Shavit" ]
null
null
We show that the error of iteratively magnitude-pruned networks empirically follows a scaling law with interpretable coefficients that depend on the architecture and task. We functionally approximate the error of the pruned networks, showing it is predictable in terms of an invariant tying width, depth, and pruning lev...
[]
null
826
2006.10621
title_snapshot
v139/ross21a
Benchmarks, Algorithms, and Metrics for Hierarchical Disentanglement
https://proceedings.mlr.press/v139/ross21a.html
[ "Andrew Ross", "Finale Doshi-Velez" ]
null
null
In representation learning, there has been recent interest in developing algorithms to disentangle the ground-truth generative factors behind a dataset, and metrics to quantify how fully this occurs. However, these algorithms and metrics often assume that both representations and ground-truth factors are flat, continuo...
[]
null
827
2102.05185
title_snapshot
v139/roth21a
Simultaneous Similarity-based Self-Distillation for Deep Metric Learning
https://proceedings.mlr.press/v139/roth21a.html
[ "Karsten Roth", "Timo Milbich", "Bjorn Ommer", "Joseph Paul Cohen", "Marzyeh Ghassemi" ]
null
null
Deep Metric Learning (DML) provides a crucial tool for visual similarity and zero-shot retrieval applications by learning generalizing embedding spaces, although recent work in DML has shown strong performance saturation across training objectives. However, generalization capacity is known to scale with the embedding s...
[]
null
828
2009.08348
title_judge
v139/rothblum21a
Multi-group Agnostic PAC Learnability
https://proceedings.mlr.press/v139/rothblum21a.html
[ "Guy N Rothblum", "Gal Yona" ]
null
null
An agnostic PAC learning algorithm finds a predictor that is competitive with the best predictor in a benchmark hypothesis class, where competitiveness is measured with respect to a given loss function. However, its predictions might be quite sub-optimal for structured subgroups of individuals, such as protected demogr...
[]
null
829
2105.09989
title_snapshot
v139/rothfuss21a
PACOH: Bayes-Optimal Meta-Learning with PAC-Guarantees
https://proceedings.mlr.press/v139/rothfuss21a.html
[ "Jonas Rothfuss", "Vincent Fortuin", "Martin Josifoski", "Andreas Krause" ]
null
null
Meta-learning can successfully acquire useful inductive biases from data. Yet, its generalization properties to unseen learning tasks are poorly understood. Particularly if the number of meta-training tasks is small, this raises concerns about overfitting. We provide a theoretical analysis using the PAC-Bayesian framew...
[]
null
830
2002.05551
title_snapshot
v139/rouyer21a
An Algorithm for Stochastic and Adversarial Bandits with Switching Costs
https://proceedings.mlr.press/v139/rouyer21a.html
[ "Chloé Rouyer", "Yevgeny Seldin", "Nicolò Cesa-Bianchi" ]
null
null
We propose an algorithm for stochastic and adversarial multiarmed bandits with switching costs, where the algorithm pays a price $\lambda$ every time it switches the arm being played. Our algorithm is based on adaptation of the Tsallis-INF algorithm of Zimmert and Seldin (2021) and requires no prior knowledge of the re...
[]
null
831
2102.09864
title_snapshot
v139/ruan21a
Improving Lossless Compression Rates via Monte Carlo Bits-Back Coding
https://proceedings.mlr.press/v139/ruan21a.html
[ "Yangjun Ruan", "Karen Ullrich", "Daniel S Severo", "James Townsend", "Ashish Khisti", "Arnaud Doucet", "Alireza Makhzani", "Chris Maddison" ]
null
null
Latent variable models have been successfully applied in lossless compression with the bits-back coding algorithm. However, bits-back suffers from an increase in the bitrate equal to the KL divergence between the approximate posterior and the true posterior. In this paper, we show how to remove this gap asymptotically ...
[]
null
832
2102.11086
title_snapshot
v139/rudner21a
On Signal-to-Noise Ratio Issues in Variational Inference for Deep Gaussian Processes
https://proceedings.mlr.press/v139/rudner21a.html
[ "Tim G. J. Rudner", "Oscar Key", "Yarin Gal", "Tom Rainforth" ]
null
null
We show that the gradient estimates used in training Deep Gaussian Processes (DGPs) with importance-weighted variational inference are susceptible to signal-to-noise ratio (SNR) issues. Specifically, we show both theoretically and via an extensive empirical evaluation that the SNR of the gradient estimates for the late...
[]
null
833
2011.00515
title_snapshot
v139/ruiz-garcia21a
Tilting the playing field: Dynamical loss functions for machine learning
https://proceedings.mlr.press/v139/ruiz-garcia21a.html
[ "Miguel Ruiz-Garcia", "Ge Zhang", "Samuel S Schoenholz", "Andrea J. Liu" ]
null
null
We show that learning can be improved by using loss functions that evolve cyclically during training to emphasize one class at a time. In underparameterized networks, such dynamical loss functions can lead to successful training for networks that fail to find deep minima of the standard cross-entropy loss. In overparam...
[]
null
834
2102.03793
title_snapshot
v139/rusch21a
UnICORNN: A recurrent model for learning very long time dependencies
https://proceedings.mlr.press/v139/rusch21a.html
[ "T. Konstantin Rusch", "Siddhartha Mishra" ]
null
null
The design of recurrent neural networks (RNNs) to accurately process sequential inputs with long-time dependencies is very challenging on account of the exploding and vanishing gradient problem. To overcome this, we propose a novel RNN architecture which is based on a structure preserving discretization of a Hamiltonia...
[]
null
835
2103.05487
title_snapshot
v139/rybkin21a
Simple and Effective VAE Training with Calibrated Decoders
https://proceedings.mlr.press/v139/rybkin21a.html
[ "Oleh Rybkin", "Kostas Daniilidis", "Sergey Levine" ]
null
null
Variational autoencoders (VAEs) provide an effective and simple method for modeling complex distributions. However, training VAEs often requires considerable hyperparameter tuning to determine the optimal amount of information retained by the latent variable. We study the impact of calibrated decoders, which learn the ...
[]
null
836
2006.13202
title_snapshot
v139/rybkin21b
Model-Based Reinforcement Learning via Latent-Space Collocation
https://proceedings.mlr.press/v139/rybkin21b.html
[ "Oleh Rybkin", "Chuning Zhu", "Anusha Nagabandi", "Kostas Daniilidis", "Igor Mordatch", "Sergey Levine" ]
null
null
The ability to plan into the future while utilizing only raw high-dimensional observations, such as images, can provide autonomous agents with broad and general capabilities. However, realistic tasks require performing temporally extended reasoning, and cannot be solved with only myopic, short-sighted planning. Recent ...
[]
null
837
2106.13229
title_snapshot
v139/s21a
Training Data Subset Selection for Regression with Controlled Generalization Error
https://proceedings.mlr.press/v139/s21a.html
[ "Durga S", "Rishabh Iyer", "Ganesh Ramakrishnan", "Abir De" ]
null
null
Data subset selection from a large number of training instances has been a successful approach toward efficient and cost-effective machine learning. However, models trained on a smaller subset may show poor generalization ability. In this paper, our goal is to design an algorithm for selecting a subset of the training ...
[]
null
838
2106.12491
title_snapshot
v139/sabour21a
Unsupervised Part Representation by Flow Capsules
https://proceedings.mlr.press/v139/sabour21a.html
[ "Sara Sabour", "Andrea Tagliasacchi", "Soroosh Yazdani", "Geoffrey Hinton", "David J Fleet" ]
null
null
Capsule networks aim to parse images into a hierarchy of objects, parts and relations. While promising, they remain limited by an inability to learn effective low level part descriptions. To address this issue we propose a way to learn primary capsule encoders that detect atomic parts from a single image. During traini...
[]
null
839
2011.13920
title_snapshot
v139/safaryan21a
Stochastic Sign Descent Methods: New Algorithms and Better Theory
https://proceedings.mlr.press/v139/safaryan21a.html
[ "Mher Safaryan", "Peter Richtarik" ]
null
null
Various gradient compression schemes have been proposed to mitigate the communication cost in distributed training of large scale machine learning models. Sign-based methods, such as signSGD (Bernstein et al., 2018), have recently been gaining popularity because of their simple compression rule and connection to adapti...
[]
null
840
1905.12938
title_snapshot
v139/saha21a
Adversarial Dueling Bandits
https://proceedings.mlr.press/v139/saha21a.html
[ "Aadirupa Saha", "Tomer Koren", "Yishay Mansour" ]
null
null
We introduce the problem of regret minimization in Adversarial Dueling Bandits. As in classic Dueling Bandits, the learner has to repeatedly choose a pair of items and observe only a relative binary ‘win-loss’ feedback for this pair, but here this feedback is generated from an arbitrary preference matrix, possibly chos...
[]
null
841
2010.14563
title_snapshot
v139/saha21b
Dueling Convex Optimization
https://proceedings.mlr.press/v139/saha21b.html
[ "Aadirupa Saha", "Tomer Koren", "Yishay Mansour" ]
null
null
We address the problem of convex optimization with preference (dueling) feedback. Like the traditional optimization objective, the goal is to find the optimal point with the least possible query complexity, however, without the luxury of even a zeroth order feedback. Instead, the learner can only observe a single noisy...
[]
null
842
null
null
v139/saha21c
Optimal regret algorithm for Pseudo-1d Bandit Convex Optimization
https://proceedings.mlr.press/v139/saha21c.html
[ "Aadirupa Saha", "Nagarajan Natarajan", "Praneeth Netrapalli", "Prateek Jain" ]
null
null
We study online learning with bandit feedback (i.e. learner has access to only zeroth-order oracle) where cost/reward functions $\f_t$ admit a "pseudo-1d" structure, i.e. $\f_t(\w) = \loss_t(\pred_t(\w))$ where the output of $\pred_t$ is one-dimensional. At each round, the learner observes context $\x_t$, plays predict...
[]
null
843
2102.07387
title_snapshot
v139/sahraee-ardakan21a
Asymptotics of Ridge Regression in Convolutional Models
https://proceedings.mlr.press/v139/sahraee-ardakan21a.html
[ "Mojtaba Sahraee-Ardakan", "Tung Mai", "Anup Rao", "Ryan A. Rossi", "Sundeep Rangan", "Alyson K Fletcher" ]
null
null
Understanding generalization and estimation error of estimators for simple models such as linear and generalized linear models has attracted a lot of attention recently. This is in part due to an interesting observation made in machine learning community that highly over-parameterized neural networks achieve zero train...
[]
null
844
2103.04557
title_snapshot
v139/sander21a
Momentum Residual Neural Networks
https://proceedings.mlr.press/v139/sander21a.html
[ "Michael E. Sander", "Pierre Ablin", "Mathieu Blondel", "Gabriel Peyré" ]
null
null
The training of deep residual neural networks (ResNets) with backpropagation has a memory cost that increases linearly with respect to the depth of the network. A simple way to circumvent this issue is to use reversible architectures. In this paper, we propose to change the forward rule of a ResNet by adding a momentum...
[]
null
845
2102.07870
title_snapshot
v139/sandler21a
Meta-Learning Bidirectional Update Rules
https://proceedings.mlr.press/v139/sandler21a.html
[ "Mark Sandler", "Max Vladymyrov", "Andrey Zhmoginov", "Nolan Miller", "Tom Madams", "Andrew Jackson", "Blaise Agüera Y Arcas" ]
null
null
In this paper, we introduce a new type of generalized neural network where neurons and synapses maintain multiple states. We show that classical gradient-based backpropagation in neural networks can be seen as a special case of a two-state network where one state is used for activations and another for gradients, with ...
[]
null
846
2104.04657
title_snapshot
v139/sarafian21a
Recomposing the Reinforcement Learning Building Blocks with Hypernetworks
https://proceedings.mlr.press/v139/sarafian21a.html
[ "Elad Sarafian", "Shai Keynan", "Sarit Kraus" ]
null
null
The Reinforcement Learning (RL) building blocks, i.e. $Q$-functions and policy networks, usually take elements from the cartesian product of two domains as input. In particular, the input of the $Q$-function is both the state and the action, and in multi-task problems (Meta-RL) the policy can take a state and a context...
[]
null
847
2106.06842
title_snapshot
v139/sarussi21a
Towards Understanding Learning in Neural Networks with Linear Teachers
https://proceedings.mlr.press/v139/sarussi21a.html
[ "Roei Sarussi", "Alon Brutzkus", "Amir Globerson" ]
null
null
Can a neural network minimizing cross-entropy learn linearly separable data? Despite progress in the theory of deep learning, this question remains unsolved. Here we prove that SGD globally optimizes this learning problem for a two-layer network with Leaky ReLU activations. The learned network can in principle be very ...
[]
null
848
2101.02533
title_snapshot
v139/satorras21a
E(n) Equivariant Graph Neural Networks
https://proceedings.mlr.press/v139/satorras21a.html
[ "Vı́ctor Garcia Satorras", "Emiel Hoogeboom", "Max Welling" ]
null
null
This paper introduces a new model to learn graph neural networks equivariant to rotations, translations, reflections and permutations called E(n)-Equivariant Graph Neural Networks (EGNNs). In contrast with existing methods, our work does not require computationally expensive higher-order representations in intermediate...
[]
null
849
2102.09844
title_snapshot
v139/saunshi21a
A Representation Learning Perspective on the Importance of Train-Validation Splitting in Meta-Learning
https://proceedings.mlr.press/v139/saunshi21a.html
[ "Nikunj Saunshi", "Arushi Gupta", "Wei Hu" ]
null
null
An effective approach in meta-learning is to utilize multiple “train tasks” to learn a good initialization for model parameters that can help solve unseen “test tasks” with very few samples by fine-tuning from this initialization. Although successful in practice, theoretical understanding of such methods is limited. Th...
[]
null
850
2106.15615
title_snapshot
v139/scetbon21a
Low-Rank Sinkhorn Factorization
https://proceedings.mlr.press/v139/scetbon21a.html
[ "Meyer Scetbon", "Marco Cuturi", "Gabriel Peyré" ]
null
null
Several recent applications of optimal transport (OT) theory to machine learning have relied on regularization, notably entropy and the Sinkhorn algorithm. Because matrix-vector products are pervasive in the Sinkhorn algorithm, several works have proposed to \textit{approximate} kernel matrices appearing in its iterati...
[]
null
851
2103.04737
title_snapshot
v139/schlag21a
Linear Transformers Are Secretly Fast Weight Programmers
https://proceedings.mlr.press/v139/schlag21a.html
[ "Imanol Schlag", "Kazuki Irie", "Jürgen Schmidhuber" ]
null
null
We show the formal equivalence of linearised self-attention mechanisms and fast weight controllers from the early ’90s, where a slow neural net learns by gradient descent to program the fast weights of another net through sequences of elementary programming instructions which are additive outer products of self-invente...
[]
null
852
2102.11174
title_snapshot
v139/schmidt21a
Descending through a Crowded Valley - Benchmarking Deep Learning Optimizers
https://proceedings.mlr.press/v139/schmidt21a.html
[ "Robin M Schmidt", "Frank Schneider", "Philipp Hennig" ]
null
null
Choosing the optimizer is considered to be among the most crucial design decisions in deep learning, and it is not an easy one. The growing literature now lists hundreds of optimization methods. In the absence of clear theoretical guidance and conclusive empirical evidence, the decision is often made based on anecdotes...
[]
null
853
2007.01547
title_snapshot
v139/schutt21a
Equivariant message passing for the prediction of tensorial properties and molecular spectra
https://proceedings.mlr.press/v139/schutt21a.html
[ "Kristof Schütt", "Oliver Unke", "Michael Gastegger" ]
null
null
Message passing neural networks have become a method of choice for learning on graphs, in particular the prediction of chemical properties and the acceleration of molecular dynamics studies. While they readily scale to large training data sets, previous approaches have proven to be less data efficient than kernel metho...
[]
null
854
2102.03150
title_snapshot
v139/schwarzschild21a
Just How Toxic is Data Poisoning? A Unified Benchmark for Backdoor and Data Poisoning Attacks
https://proceedings.mlr.press/v139/schwarzschild21a.html
[ "Avi Schwarzschild", "Micah Goldblum", "Arjun Gupta", "John P Dickerson", "Tom Goldstein" ]
null
null
Data poisoning and backdoor attacks manipulate training data in order to cause models to fail during inference. A recent survey of industry practitioners found that data poisoning is the number one concern among threats ranging from model stealing to adversarial attacks. However, it remains unclear exactly how dangerou...
[]
null
855
2006.12557
title_snapshot
v139/scieur21a
Connecting Sphere Manifolds Hierarchically for Regularization
https://proceedings.mlr.press/v139/scieur21a.html
[ "Damien Scieur", "Youngsung Kim" ]
null
null
This paper considers classification problems with hierarchically organized classes. We force the classifier (hyperplane) of each class to belong to a sphere manifold, whose center is the classifier of its super-class. Then, individual sphere manifolds are connected based on their hierarchical relations. Our technique r...
[]
null
856
2106.13549
title_snapshot
v139/seidenschwarz21a
Learning Intra-Batch Connections for Deep Metric Learning
https://proceedings.mlr.press/v139/seidenschwarz21a.html
[ "Jenny Denise Seidenschwarz", "Ismail Elezi", "Laura Leal-Taixé" ]
null
null
The goal of metric learning is to learn a function that maps samples to a lower-dimensional space where similar samples lie closer than dissimilar ones. Particularly, deep metric learning utilizes neural networks to learn such a mapping. Most approaches rely on losses that only take the relations between pairs or tripl...
[]
null
857
2102.07753
title_snapshot
v139/sen21a
Top-k eXtreme Contextual Bandits with Arm Hierarchy
https://proceedings.mlr.press/v139/sen21a.html
[ "Rajat Sen", "Alexander Rakhlin", "Lexing Ying", "Rahul Kidambi", "Dean Foster", "Daniel N Hill", "Inderjit S. Dhillon" ]
null
null
Motivated by modern applications, such as online advertisement and recommender systems, we study the top-$k$ extreme contextual bandits problem, where the total number of arms can be enormous, and the learner is allowed to select $k$ arms and observe all or some of the rewards for the chosen arms. We first propose an a...
[]
null
858
2102.07800
title_snapshot
v139/sentenac21a
Pure Exploration and Regret Minimization in Matching Bandits
https://proceedings.mlr.press/v139/sentenac21a.html
[ "Flore Sentenac", "Jialin Yi", "Clement Calauzenes", "Vianney Perchet", "Milan Vojnovic" ]
null
null
Finding an optimal matching in a weighted graph is a standard combinatorial problem. We consider its semi-bandit version where either a pair or a full matching is sampled sequentially. We prove that it is possible to leverage a rank-1 assumption on the adjacency matrix to reduce the sample complexity and the regret of ...
[]
null
859
2108.00230
title_snapshot
v139/seo21a
State Entropy Maximization with Random Encoders for Efficient Exploration
https://proceedings.mlr.press/v139/seo21a.html
[ "Younggyo Seo", "Lili Chen", "Jinwoo Shin", "Honglak Lee", "Pieter Abbeel", "Kimin Lee" ]
null
null
Recent exploration methods have proven to be a recipe for improving sample-efficiency in deep reinforcement learning (RL). However, efficient exploration in high-dimensional observation spaces still remains a challenge. This paper presents Random Encoders for Efficient Exploration (RE3), an exploration method that util...
[]
null
860
2102.09430
title_snapshot
v139/sessa21a
Online Submodular Resource Allocation with Applications to Rebalancing Shared Mobility Systems
https://proceedings.mlr.press/v139/sessa21a.html
[ "Pier Giuseppe Sessa", "Ilija Bogunovic", "Andreas Krause", "Maryam Kamgarpour" ]
null
null
Motivated by applications in shared mobility, we address the problem of allocating a group of agents to a set of resources to maximize a cumulative welfare objective. We model the welfare obtainable from each resource as a monotone DR-submodular function which is a-priori unknown and can only be learned by observing th...
[]
null
861
null
null
v139/shah21a
RRL: Resnet as representation for Reinforcement Learning
https://proceedings.mlr.press/v139/shah21a.html
[ "Rutav M Shah", "Vikash Kumar" ]
null
null
The ability to autonomously learn behaviors via direct interactions in uninstrumented environments can lead to generalist robots capable of enhancing productivity or providing care in unstructured settings like homes. Such uninstrumented settings warrant operations only using the robot’s proprioceptive sensor such as o...
[]
null
862
2107.03380
title_snapshot
v139/shakerinava21a
Equivariant Networks for Pixelized Spheres
https://proceedings.mlr.press/v139/shakerinava21a.html
[ "Mehran Shakerinava", "Siamak Ravanbakhsh" ]
null
null
Pixelizations of Platonic solids such as the cube and icosahedron have been widely used to represent spherical data, from climate records to Cosmic Microwave Background maps. Platonic solids have well-known global symmetries. Once we pixelize each face of the solid, each face also possesses its own local symmetries in ...
[]
null
863
2106.06662
title_snapshot
v139/shamsian21a
Personalized Federated Learning using Hypernetworks
https://proceedings.mlr.press/v139/shamsian21a.html
[ "Aviv Shamsian", "Aviv Navon", "Ethan Fetaya", "Gal Chechik" ]
null
null
Personalized federated learning is tasked with training machine learning models for multiple clients, each with its own data distribution. The goal is to train personalized models collaboratively while accounting for data disparities across clients and reducing communication costs. We propose a novel approach to this p...
[]
null
864
2103.04628
title_snapshot
v139/shen21a
On the Power of Localized Perceptron for Label-Optimal Learning of Halfspaces with Adversarial Noise
https://proceedings.mlr.press/v139/shen21a.html
[ "Jie Shen" ]
null
null
We study {\em online} active learning of homogeneous halfspaces in $\mathbb{R}^d$ with adversarial noise where the overall probability of a noisy label is constrained to be at most $\nu$. Our main contribution is a Perceptron-like online active learning algorithm that runs in polynomial time, and under the conditions t...
[]
null
865
2012.10793
title_snapshot
v139/shen21b
Sample-Optimal PAC Learning of Halfspaces with Malicious Noise
https://proceedings.mlr.press/v139/shen21b.html
[ "Jie Shen" ]
null
null
We study efficient PAC learning of homogeneous halfspaces in $\mathbb{R}^d$ in the presence of malicious noise of Valiant (1985). This is a challenging noise model and only until recently has near-optimal noise tolerance bound been established under the mild condition that the unlabeled data distribution is isotropic l...
[]
null
866
2102.06247
title_snapshot
v139/shen21c
Backdoor Scanning for Deep Neural Networks through K-Arm Optimization
https://proceedings.mlr.press/v139/shen21c.html
[ "Guangyu Shen", "Yingqi Liu", "Guanhong Tao", "Shengwei An", "Qiuling Xu", "Siyuan Cheng", "Shiqing Ma", "Xiangyu Zhang" ]
null
null
Back-door attack poses a severe threat to deep learning systems. It injects hidden malicious behaviors to a model such that any input stamped with a special pattern can trigger such behaviors. Detecting back-door is hence of pressing need. Many existing defense techniques use optimization to generate the smallest input...
[]
null
867
2102.05123
title_snapshot
v139/shen21d
State Relevance for Off-Policy Evaluation
https://proceedings.mlr.press/v139/shen21d.html
[ "Simon P Shen", "Yecheng Ma", "Omer Gottesman", "Finale Doshi-Velez" ]
null
null
Importance sampling-based estimators for off-policy evaluation (OPE) are valued for their simplicity, unbiasedness, and reliance on relatively few assumptions. However, the variance of these estimators is often high, especially when trajectories are of different lengths. In this work, we introduce Omitting-States-Irrel...
[]
null
868
2109.06310
title_snapshot
v139/shi21a
SparseBERT: Rethinking the Importance Analysis in Self-attention
https://proceedings.mlr.press/v139/shi21a.html
[ "Han Shi", "Jiahui Gao", "Xiaozhe Ren", "Hang Xu", "Xiaodan Liang", "Zhenguo Li", "James Tin-Yau Kwok" ]
null
null
Transformer-based models are popularly used in natural language processing (NLP). Its core component, self-attention, has aroused widespread interest. To understand the self-attention mechanism, a direct method is to visualize the attention map of a pre-trained model. Based on the patterns observed, a series of efficie...
[]
null
869
2102.12871
title_snapshot
v139/shi21b
Learning Gradient Fields for Molecular Conformation Generation
https://proceedings.mlr.press/v139/shi21b.html
[ "Chence Shi", "Shitong Luo", "Minkai Xu", "Jian Tang" ]
null
null
We study a fundamental problem in computational chemistry known as molecular conformation generation, trying to predict stable 3D structures from 2D molecular graphs. Existing machine learning approaches usually first predict distances between atoms and then generate a 3D structure satisfying the distances, where noise...
[]
null
870
2105.03902
title_snapshot
v139/shi21c
Segmenting Hybrid Trajectories using Latent ODEs
https://proceedings.mlr.press/v139/shi21c.html
[ "Ruian Shi", "Quaid Morris" ]
null
null
Smooth dynamics interrupted by discontinuities are known as hybrid systems and arise commonly in nature. Latent ODEs allow for powerful representation of irregularly sampled time series but are not designed to capture trajectories arising from hybrid systems. Here, we propose the Latent Segmented ODE (LatSegODE), which...
[]
null
871
2105.03835
title_snapshot
v139/shi21d
Deeply-Debiased Off-Policy Interval Estimation
https://proceedings.mlr.press/v139/shi21d.html
[ "Chengchun Shi", "Runzhe Wan", "Victor Chernozhukov", "Rui Song" ]
null
null
Off-policy evaluation learns a target policy’s value with a historical dataset generated by a different behavior policy. In addition to a point estimate, many applications would benefit significantly from having a confidence interval (CI) that quantifies the uncertainty of the point estimate. In this paper, we propose ...
[]
null
872
2105.04646
title_snapshot
v139/shih21a
GANMEX: One-vs-One Attributions using GAN-based Model Explainability
https://proceedings.mlr.press/v139/shih21a.html
[ "Sheng-Min Shih", "Pin-Ju Tien", "Zohar Karnin" ]
null
null
Attribution methods have been shown as promising approaches for identifying key features that led to learned model predictions. While most existing attribution methods rely on a baseline input for performing feature perturbations, limited research has been conducted to address the baseline selection issues. Poor choice...
[]
null
873
null
null
v139/shin21a
Large-Scale Meta-Learning with Continual Trajectory Shifting
https://proceedings.mlr.press/v139/shin21a.html
[ "Jaewoong Shin", "Hae Beom Lee", "Boqing Gong", "Sung Ju Hwang" ]
null
null
Meta-learning of shared initialization parameters has shown to be highly effective in solving few-shot learning tasks. However, extending the framework to many-shot scenarios, which may further enhance its practicality, has been relatively overlooked due to the technical difficulties of meta-learning over long chains o...
[]
null
874
2102.07215
title_snapshot
v139/shu21a
AGENT: A Benchmark for Core Psychological Reasoning
https://proceedings.mlr.press/v139/shu21a.html
[ "Tianmin Shu", "Abhishek Bhandwaldar", "Chuang Gan", "Kevin Smith", "Shari Liu", "Dan Gutfreund", "Elizabeth Spelke", "Joshua Tenenbaum", "Tomer Ullman" ]
null
null
For machine agents to successfully interact with humans in real-world settings, they will need to develop an understanding of human mental life. Intuitive psychology, the ability to reason about hidden mental variables that drive observable actions, comes naturally to people: even pre-verbal infants can tell agents fro...
[]
null
875
2102.12321
title_snapshot
v139/shu21b
Zoo-Tuning: Adaptive Transfer from A Zoo of Models
https://proceedings.mlr.press/v139/shu21b.html
[ "Yang Shu", "Zhi Kou", "Zhangjie Cao", "Jianmin Wang", "Mingsheng Long" ]
null
null
With the development of deep networks on various large-scale datasets, a large zoo of pretrained models are available. When transferring from a model zoo, applying classic single-model-based transfer learning methods to each source model suffers from high computational cost and cannot fully utilize the rich knowledge i...
[]
null
876
2106.15434
title_snapshot
v139/shui21a
Aggregating From Multiple Target-Shifted Sources
https://proceedings.mlr.press/v139/shui21a.html
[ "Changjian Shui", "Zijian Li", "Jiaqi Li", "Christian Gagné", "Charles X Ling", "Boyu Wang" ]
null
null
Multi-source domain adaptation aims at leveraging the knowledge from multiple tasks for predicting a related target domain. Hence, a crucial aspect is to properly combine different sources based on their relations. In this paper, we analyzed the problem for aggregating source domains with different label distributions,...
[]
null
877
2105.04051
title_snapshot
v139/si21a
Testing Group Fairness via Optimal Transport Projections
https://proceedings.mlr.press/v139/si21a.html
[ "Nian Si", "Karthyek Murthy", "Jose Blanchet", "Viet Anh Nguyen" ]
null
null
We have developed a statistical testing framework to detect if a given machine learning classifier fails to satisfy a wide range of group fairness notions. Our test is a flexible, interpretable, and statistically rigorous tool for auditing whether exhibited biases are intrinsic to the algorithm or simply due to the ran...
[]
null
878
2106.01070
title_snapshot
v139/sidheekh21a
On Characterizing GAN Convergence Through Proximal Duality Gap
https://proceedings.mlr.press/v139/sidheekh21a.html
[ "Sahil Sidheekh", "Aroof Aimen", "Narayanan C Krishnan" ]
null
null
Despite the accomplishments of Generative Adversarial Networks (GANs) in modeling data distributions, training them remains a challenging task. A contributing factor to this difficulty is the non-intuitive nature of the GAN loss curves, which necessitates a subjective evaluation of the generated output to infer trainin...
[]
null
879
2105.04801
title_snapshot
v139/sifaou21a
A Precise Performance Analysis of Support Vector Regression
https://proceedings.mlr.press/v139/sifaou21a.html
[ "Houssem Sifaou", "Abla Kammoun", "Mohamed-Slim Alouini" ]
null
null
In this paper, we study the hard and soft support vector regression techniques applied to a set of $n$ linear measurements of the form $y_i=\boldsymbol{\beta}_\star^{T}{\bf x}_i +n_i$ where $\boldsymbol{\beta}_\star$ is an unknown vector, $\left\{{\bf x}_i\right\}_{i=1}^n$ are the feature vectors and $\left\{{n}_i\righ...
[]
null
880
2105.10373
title_snapshot
v139/sim21a
Directed Graph Embeddings in Pseudo-Riemannian Manifolds
https://proceedings.mlr.press/v139/sim21a.html
[ "Aaron Sim", "Maciej L Wiatrak", "Angus Brayne", "Paidi Creed", "Saee Paliwal" ]
null
null
The inductive biases of graph representation learning algorithms are often encoded in the background geometry of their embedding space. In this paper, we show that general directed graphs can be effectively represented by an embedding model that combines three components: a pseudo-Riemannian metric structure, a non-tri...
[]
null
881
2106.08678
title_snapshot
v139/sim21b
Collaborative Bayesian Optimization with Fair Regret
https://proceedings.mlr.press/v139/sim21b.html
[ "Rachael Hwee Ling Sim", "Yehong Zhang", "Bryan Kian Hsiang Low", "Patrick Jaillet" ]
null
null
Bayesian optimization (BO) is a popular tool for optimizing complex and costly-to-evaluate black-box objective functions. To further reduce the number of function evaluations, any party performing BO may be interested to collaborate with others to optimize the same objective function concurrently. To do this, existing ...
[]
null
882
null
null
v139/simchi-levi21a
Dynamic Planning and Learning under Recovering Rewards
https://proceedings.mlr.press/v139/simchi-levi21a.html
[ "David Simchi-Levi", "Zeyu Zheng", "Feng Zhu" ]
null
null
Motivated by emerging applications such as live-streaming e-commerce, promotions and recommendations, we introduce a general class of multi-armed bandit problems that have the following two features: (i) the decision maker can pull and collect rewards from at most $K$ out of $N$ different arms in each time period; (ii)...
[]
null
883
null
null
v139/simon-gabriel21a
PopSkipJump: Decision-Based Attack for Probabilistic Classifiers
https://proceedings.mlr.press/v139/simon-gabriel21a.html
[ "Carl-Johann Simon-Gabriel", "Noman Ahmed Sheikh", "Andreas Krause" ]
null
null
Most current classifiers are vulnerable to adversarial examples, small input perturbations that change the classification output. Many existing attack algorithms cover various settings, from white-box to black-box classifiers, but usually assume that the answers are deterministic and often fail when they are not. We th...
[]
null
884
2106.07445
title_snapshot
v139/simsek21a
Geometry of the Loss Landscape in Overparameterized Neural Networks: Symmetries and Invariances
https://proceedings.mlr.press/v139/simsek21a.html
[ "Berfin Simsek", "François Ged", "Arthur Jacot", "Francesco Spadaro", "Clement Hongler", "Wulfram Gerstner", "Johanni Brea" ]
null
null
We study how permutation symmetries in overparameterized multi-layer neural networks generate ‘symmetry-induced’ critical points. Assuming a network with $ L $ layers of minimal widths $ r_1^*, \ldots, r_{L-1}^* $ reaches a zero-loss minimum at $ r_1^*! \cdots r_{L-1}^*! $ isolated points that are permutations of one a...
[]
null
885
2105.12221
title_snapshot
v139/singal21a
Flow-based Attribution in Graphical Models: A Recursive Shapley Approach
https://proceedings.mlr.press/v139/singal21a.html
[ "Raghav Singal", "George Michailidis", "Hoiyi Ng" ]
null
null
We study the attribution problem in a graphical model, wherein the objective is to quantify how the effect of changes at the source nodes propagates through the graph. We develop a model-agnostic flow-based attribution method, called recursive Shapley value (RSV). RSV generalizes a number of existing node-based methods...
[]
null
886
null
null
v139/singh21a
Structured World Belief for Reinforcement Learning in POMDP
https://proceedings.mlr.press/v139/singh21a.html
[ "Gautam Singh", "Skand Peri", "Junghyun Kim", "Hyunseok Kim", "Sungjin Ahn" ]
null
null
Object-centric world models provide structured representation of the scene and can be an important backbone in reinforcement learning and planning. However, existing approaches suffer in partially-observable environments due to the lack of belief states. In this paper, we propose Structured World Belief, a model for le...
[]
null
887
2107.08577
title_snapshot
v139/singla21a
Skew Orthogonal Convolutions
https://proceedings.mlr.press/v139/singla21a.html
[ "Sahil Singla", "Soheil Feizi" ]
null
null
Training convolutional neural networks with a Lipschitz constraint under the $l_{2}$ norm is useful for provable adversarial robustness, interpretable gradients, stable training, etc. While 1-Lipschitz networks can be designed by imposing a 1-Lipschitz constraint on each layer, training such networks requires each laye...
[]
null
888
2105.11417
title_snapshot
v139/sodhani21a
Multi-Task Reinforcement Learning with Context-based Representations
https://proceedings.mlr.press/v139/sodhani21a.html
[ "Shagun Sodhani", "Amy Zhang", "Joelle Pineau" ]
null
null
https://drive.google.com/file/d/1lRV72XaKoxZjgQrLXBJhsM82x54_1Vc4/view?usp=sharing
[]
null
889
2102.06177
title_snapshot
v139/sohn21a
Shortest-Path Constrained Reinforcement Learning for Sparse Reward Tasks
https://proceedings.mlr.press/v139/sohn21a.html
[ "Sungryull Sohn", "Sungtae Lee", "Jongwook Choi", "Harm H Van Seijen", "Mehdi Fatemi", "Honglak Lee" ]
null
null
We propose the k-Shortest-Path (k-SP) constraint: a novel constraint on the agent’s trajectory that improves the sample efficiency in sparse-reward MDPs. We show that any optimal policy necessarily satisfies the k-SP constraint. Notably, the k-SP constraint prevents the policy from exploring state-action pairs along th...
[]
null
890
2107.06405
title_snapshot
v139/song21a
Accelerating Feedforward Computation via Parallel Nonlinear Equation Solving
https://proceedings.mlr.press/v139/song21a.html
[ "Yang Song", "Chenlin Meng", "Renjie Liao", "Stefano Ermon" ]
null
null
Feedforward computation, such as evaluating a neural network or sampling from an autoregressive model, is ubiquitous in machine learning. The sequential nature of feedforward computation, however, requires a strict order of execution and cannot be easily accelerated with parallel computing. To enable parallelization, w...
[]
null
891
2002.03629
title_snapshot
v139/song21b
PC-MLP: Model-based Reinforcement Learning with Policy Cover Guided Exploration
https://proceedings.mlr.press/v139/song21b.html
[ "Yuda Song", "Wen Sun" ]
null
null
Model-based Reinforcement Learning (RL) is a popular learning paradigm due to its potential sample efficiency compared to model-free RL. However, existing empirical model-based RL approaches lack the ability to explore. This work studies a computationally and statistically efficient model-based algorithm for both Kerne...
[]
null
892
2107.07410
title_snapshot
v139/song21c
Fast Sketching of Polynomial Kernels of Polynomial Degree
https://proceedings.mlr.press/v139/song21c.html
[ "Zhao Song", "David Woodruff", "Zheng Yu", "Lichen Zhang" ]
null
null
Kernel methods are fundamental in machine learning, and faster algorithms for kernel approximation provide direct speedups for many core tasks in machine learning. The polynomial kernel is especially important as other kernels can often be approximated by the polynomial kernel via a Taylor series expansion. Recent tech...
[]
null
893
2108.09420
title_snapshot
v139/song21d
Variance Reduction via Primal-Dual Accelerated Dual Averaging for Nonsmooth Convex Finite-Sums
https://proceedings.mlr.press/v139/song21d.html
[ "Chaobing Song", "Stephen J Wright", "Jelena Diakonikolas" ]
null
null
Structured nonsmooth convex finite-sum optimization appears in many machine learning applications, including support vector machines and least absolute deviation. For the primal-dual formulation of this problem, we propose a novel algorithm called \emph{Variance Reduction via Primal-Dual Accelerated Dual Averaging (\vr...
[]
null
894
2102.13643
title_snapshot
v139/song21e
Oblivious Sketching-based Central Path Method for Linear Programming
https://proceedings.mlr.press/v139/song21e.html
[ "Zhao Song", "Zheng Yu" ]
null
null
In this work, we propose a sketching-based central path method for solving linear programmings, whose running time matches the state of the art results [Cohen, Lee, Song STOC 19; Lee, Song, Zhang COLT 19]. Our method opens up the iterations of the central path method and deploys an "iterate and sketch" approach towards...
[]
null
895
null
null
v139/sontakke21a
Causal Curiosity: RL Agents Discovering Self-supervised Experiments for Causal Representation Learning
https://proceedings.mlr.press/v139/sontakke21a.html
[ "Sumedh A Sontakke", "Arash Mehrjou", "Laurent Itti", "Bernhard Schölkopf" ]
null
null
Humans show an innate ability to learn the regularities of the world through interaction. By performing experiments in our environment, we are able to discern the causal factors of variation and infer how they affect the dynamics of our world. Analogously, here we attempt to equip reinforcement learning agents with the...
[]
null
896
2010.03110
title_snapshot
v139/sordoni21a
Decomposed Mutual Information Estimation for Contrastive Representation Learning
https://proceedings.mlr.press/v139/sordoni21a.html
[ "Alessandro Sordoni", "Nouha Dziri", "Hannes Schulz", "Geoff Gordon", "Philip Bachman", "Remi Tachet Des Combes" ]
null
null
Recent contrastive representation learning methods rely on estimating mutual information (MI) between multiple views of an underlying context. E.g., we can derive multiple views of a given image by applying data augmentation, or we can split a sequence into views comprising the past and future of some step in the seque...
[]
null
897
2106.13401
title_snapshot
v139/stooke21a
Decoupling Representation Learning from Reinforcement Learning
https://proceedings.mlr.press/v139/stooke21a.html
[ "Adam Stooke", "Kimin Lee", "Pieter Abbeel", "Michael Laskin" ]
null
null
In an effort to overcome limitations of reward-driven feature learning in deep reinforcement learning (RL) from images, we propose decoupling representation learning from policy learning. To this end, we introduce a new unsupervised learning (UL) task, called Augmented Temporal Contrast (ATC), which trains a convolutio...
[]
null
898
2009.08319
title_snapshot
v139/su21a
K-shot NAS: Learnable Weight-Sharing for NAS with K-shot Supernets
https://proceedings.mlr.press/v139/su21a.html
[ "Xiu Su", "Shan You", "Mingkai Zheng", "Fei Wang", "Chen Qian", "Changshui Zhang", "Chang Xu" ]
null
null
In one-shot weight sharing for NAS, the weights of each operation (at each layer) are supposed to be identical for all architectures (paths) in the supernet. However, this rules out the possibility of adjusting operation weights to cater for different paths, which limits the reliability of the evaluation results. In th...
[]
null
899
2106.06442
title_snapshot
v139/sugiyama21a
More Powerful and General Selective Inference for Stepwise Feature Selection using Homotopy Method
https://proceedings.mlr.press/v139/sugiyama21a.html
[ "Kazuya Sugiyama", "Vo Nguyen Le Duy", "Ichiro Takeuchi" ]
null
null
Conditional selective inference (SI) has been actively studied as a new statistical inference framework for data-driven hypotheses. The basic idea of conditional SI is to make inferences conditional on the selection event characterized by a set of linear and/or quadratic inequalities. Conditional SI has been mainly stu...
[]
null
900
2012.13545
title_judge