title
stringlengths
12
151
paper_url
stringlengths
41
43
authors
listlengths
1
40
type
stringclasses
3 values
primary_area
stringclasses
0 values
abstract
large_stringlengths
519
2.34k
keywords
listlengths
0
19
TL;DR
large_stringclasses
0 values
submission_number
int64
1
4.72k
arxiv_id
stringlengths
10
10
arxiv_id_source
stringclasses
2 values
embedding
listlengths
768
768
Wiring Up Vision: Minimizing Supervised Synaptic Updates Needed to Produce a Primate Ventral Stream
https://openreview.net/forum?id=g1SzIRLQXMM
[ "Franziska Geiger", "Martin Schrimpf", "Tiago Marques", "James J. DiCarlo" ]
Spotlight
null
After training on large datasets, certain deep neural networks are surprisingly good models of the neural mechanisms of adult primate visual object recognition. Nevertheless, these models are considered poor models of the development of the visual system because they posit millions of sequential, precisely coordinated ...
[ "computational neuroscience", "primate visual ventral stream", "convolutional neural networks", "biologically plausible learning" ]
null
4,724
null
null
[ -0.01764817349612713, -0.0029178273398429155, -0.02592240460216999, 0.04119005426764488, 0.04466262087225914, 0.028577322140336037, 0.0442272424697876, 0.02806803584098816, -0.07039349526166916, -0.04947104677557945, -0.023302601650357246, 0.008820520713925362, -0.09648112952709198, -0.006...
Learning to Downsample for Segmentation of Ultra-High Resolution Images
https://openreview.net/forum?id=HndgQudNb91
[ "Chen Jin", "Ryutaro Tanno", "Thomy Mertzanidou", "Eleftheria Panagiotaki", "Daniel C. Alexander" ]
Poster
null
Many computer vision systems require low-cost segmentation algorithms based on deep learning, either because of the enormous size of input images or limited computational budget. Common solutions uniformly downsample the input images to meet memory constraints, assuming all pixels are equally informative. In this work,...
[ "ultra-high resolution image segmentation", "non-uniform dowmsampling", "efficient segmentation", "large volume image segmentation", "medical image segmentation" ]
null
4,722
2109.11071
title_snapshot
[ 0.021435018628835678, -0.020951498299837112, 0.007348465733230114, 0.01612038165330887, 0.033655859529972076, 0.02880050241947174, 0.008444135077297688, 0.01428818330168724, -0.016897477209568024, -0.06111558899283409, -0.01661193184554577, -0.017538953572511673, -0.06625138223171234, 0.01...
Variational Neural Cellular Automata
https://openreview.net/forum?id=7fFO4cMBx_9
[ "Rasmus Berg Palm", "Miguel González Duque", "Shyam Sudhakaran", "Sebastian Risi" ]
Poster
null
In nature, the process of cellular growth and differentiation has lead to an amazing diversity of organisms --- algae, starfish, giant sequoia, tardigrades, and orcas are all created by the same generative process. Inspired by the incredible diversity of this biological generative process, we propose a generative model...
[ "Neural Cellular Automata", "Cellular Automata", "Self-Organization", "Generative Models" ]
null
4,721
2201.12360
title_snapshot
[ -0.005360063631087542, -0.011039217002689838, -0.0235077366232872, 0.04784628376364708, 0.05615873262286186, 0.04237756505608559, 0.01619027554988861, 0.01944875717163086, -0.05705038830637932, -0.04397052899003029, 0.009604637511074543, -0.019976304844021797, -0.0701095387339592, 0.030959...
Wish you were here: Hindsight Goal Selection for long-horizon dexterous manipulation
https://openreview.net/forum?id=FKp8-pIRo3y
[ "Todor Davchev", "Oleg Olegovich Sushkov", "Jean-Baptiste Regli", "Stefan Schaal", "Yusuf Aytar", "Markus Wulfmeier", "Jon Scholz" ]
Poster
null
Complex sequential tasks in continuous-control settings often require agents to successfully traverse a set of ``narrow passages'' in their state space. Solving such tasks with a sparse reward in a sample-efficient manner poses a challenge to modern reinforcement learning (RL) due to the associated long-horizon nature ...
[ "goal-conditioned reinforcement learning", "learning from demonstrations", "long-horizon dexterous manipulation", "bi-manual manipulation" ]
null
4,719
2112.00597
title_snapshot
[ -0.050316520035266876, -0.032205164432525635, -0.010615468956530094, 0.010886461474001408, 0.06958018243312836, 0.03942965343594551, 0.01392416749149561, 0.006289495620876551, -0.06015239655971527, -0.03347127139568329, -0.02016485668718815, 0.03113621473312378, -0.03849486634135246, -0.01...
L0-Sparse Canonical Correlation Analysis
https://openreview.net/forum?id=KntaNRo6R48
[ "Ofir Lindenbaum", "Moshe Salhov", "Amir Averbuch", "Yuval Kluger" ]
Poster
null
Canonical Correlation Analysis (CCA) models are powerful for studying the associations between two sets of variables. The canonically correlated representations, termed \textit{canonical variates} are widely used in unsupervised learning to analyze unlabeled multi-modal registered datasets. Despite their success, CCA m...
[]
null
4,717
null
null
[ 0.018026307225227356, -0.015658985823392868, -0.00008224917110055685, 0.01639857515692711, 0.05151583254337311, 0.04780999943614006, 0.02783195674419403, 0.03393661975860596, -0.02372054196894169, -0.03406205028295517, -0.001812935108318925, -0.012318683788180351, -0.08210859447717667, 0.0...
Recycling Model Updates in Federated Learning: Are Gradient Subspaces Low-Rank?
https://openreview.net/forum?id=B7ZbqNLDn-_
[ "Sheikh Shams Azam", "Seyyedali Hosseinalipour", "Qiang Qiu", "Christopher Brinton" ]
Poster
null
In this paper, we question the rationale behind propagating large numbers of parameters through a distributed system during federated learning. We start by examining the rank characteristics of the subspace spanned by gradients (i.e., the gradient-space) in centralized model training, and observe that the gradient-spac...
[ "Distributed Machine Learning", "Federated Learning", "Gradient Subspace", "SGD" ]
null
4,715
2202.00280
title_snapshot
[ -0.015477419830858707, -0.06269529461860657, 0.013516608625650406, 0.04058767110109329, 0.03592118248343468, 0.00046682069660164416, 0.030264882370829582, -0.005545417312532663, -0.022280702367424965, -0.0405130572617054, -0.0016693559009581804, 0.013270589523017406, -0.06704633682966232, ...
Is Homophily a Necessity for Graph Neural Networks?
https://openreview.net/forum?id=ucASPPD9GKN
[ "Yao Ma", "Xiaorui Liu", "Neil Shah", "Jiliang Tang" ]
Poster
null
Graph neural networks (GNNs) have shown great prowess in learning representations suitable for numerous graph-based machine learning tasks. When applied to semi-supervised node classification, GNNs are widely believed to work well due to the homophily assumption (``like attracts like''), and fail to generalize to hete...
[]
null
4,711
2106.06134
title_snapshot
[ 0.0026286013890057802, -0.02790139615535736, -0.00008399521902902052, 0.04326372593641281, 0.030910884961485863, 0.020450923591852188, 0.011659958399832249, 0.0167634766548872, -0.007178389001637697, -0.03309507668018341, 0.012980584055185318, -0.021431654691696167, -0.08292808383703232, 0...
DEGREE: Decomposition Based Explanation for Graph Neural Networks
https://openreview.net/forum?id=Ve0Wth3ptT_
[ "Qizhang Feng", "Ninghao Liu", "Fan Yang", "Ruixiang Tang", "Mengnan Du", "Xia Hu" ]
Poster
null
Graph Neural Networks (GNNs) are gaining extensive attention for their application in graph data. However, the black-box nature of GNNs prevents users from understanding and trusting the models, thus hampering their applicability. Whereas explaining GNNs remains a challenge, most existing methods fall into approximatio...
[ "XAI", "GNN" ]
null
4,703
2305.12895
title_snapshot
[ -0.023520521819591522, 0.0030457451939582825, -0.007767949253320694, 0.04722379520535469, 0.03821787238121033, 0.048065610229969025, 0.01561952754855156, 0.0038703165482729673, -0.040815770626068115, -0.029920615255832672, 0.018120290711522102, -0.000737907481379807, -0.07385191321372986, ...
Improving Mutual Information Estimation with Annealed and Energy-Based Bounds
https://openreview.net/forum?id=T0B9AoM_bFg
[ "Rob Brekelmans", "Sicong Huang", "Marzyeh Ghassemi", "Greg Ver Steeg", "Roger Baker Grosse", "Alireza Makhzani" ]
Poster
null
Mutual information (MI) is a fundamental quantity in information theory and machine learning. However, direct estimation of MI is intractable, even if the true joint probability density for the variables of interest is known, as it involves estimating a potentially high-dimensional log partition function. In this work,...
[ "mutual information estimation", "annealed importance sampling", "energy-based models" ]
null
4,668
2303.06992
title_snapshot
[ -0.020771825686097145, 0.005026770755648613, -0.011725908145308495, 0.05385679751634598, 0.030470537021756172, 0.01442105695605278, 0.03916197270154953, -0.021159835159778595, -0.01371364388614893, -0.0487983375787735, 0.006430924404412508, 0.006725159008055925, -0.08481135219335556, -0.01...
Sequence Approximation using Feedforward Spiking Neural Network for Spatiotemporal Learning: Theory and Optimization Methods
https://openreview.net/forum?id=bp-LJ4y_XC
[ "Xueyuan She", "Saurabh Dash", "Saibal Mukhopadhyay" ]
Poster
null
A dynamical system of spiking neurons with only feedforward connections can classify spatiotemporal patterns without recurrent connections. However, the theoretical construct of a feedforward spiking neural network (SNN) for approximating a temporal sequence remains unclear, making it challenging to optimize SNN archit...
[ "spiking neural network", "spatiotemporal processing", "feedforward network" ]
null
4,662
null
null
[ -0.01456733513623476, -0.0022099989000707865, 0.013063693419098854, 0.012743940576910973, 0.03830643370747566, 0.025833213701844215, 0.029463794082403183, 0.021953968331217766, -0.05756936967372894, -0.02232467755675316, 0.01556016132235527, -0.035320959985256195, -0.03676305711269379, 0.0...
Diverse Client Selection for Federated Learning via Submodular Maximization
https://openreview.net/forum?id=nwKXyFvaUm
[ "Ravikumar Balakrishnan", "Tian Li", "Tianyi Zhou", "Nageen Himayat", "Virginia Smith", "Jeff Bilmes" ]
Poster
null
In every communication round of federated learning, a random subset of clients communicate their model updates back to the server which then aggregates them all. The optimal size of this subset is not known and several studies have shown that typically random selection does not perform very well in terms of ...
[ "federated learning", "submodularity", "diversity" ]
null
4,660
null
null
[ -0.01380859687924385, -0.08271554112434387, 0.004783566575497389, 0.05112672969698906, 0.042195875197649, 0.03563714399933815, 0.01719977892935276, -0.004018982872366905, -0.020217290148139, -0.049162253737449646, 0.0026308875530958176, -0.002781166695058346, -0.07265185564756393, -0.00030...
From Intervention to Domain Transportation: A Novel Perspective to Optimize Recommendation
https://openreview.net/forum?id=jT1EwXu-4hj
[ "Da Xu", "Yuting Ye", "Chuanwei Ruan", "Evren Korpeoglu", "Sushant Kumar", "Kannan Achan" ]
Poster
null
The interventional nature of recommendation has attracted increasing attention in recent years. It particularly motivates researchers to formulate learning and evaluating recommendation as causal inference and data missing-not-at-random problems. However, few take seriously the consequence of violating the critical ass...
[ "Information retrieval", "Learning theory", "Causal inference", "Missing data", "Overlapping", "Reweighting", "Optimal transport" ]
null
4,651
2203.13956
title_snapshot
[ -0.03320901095867157, -0.035220853984355927, 0.01708921603858471, 0.052199240773916245, 0.06618726253509521, -0.030693061649799347, 0.027237029746174812, -0.014162295497953892, -0.010777859017252922, -0.05014178156852722, -0.03588714450597763, 0.034188613295555115, -0.051513005048036575, 0...
Variational Predictive Routing with Nested Subjective Timescales
https://openreview.net/forum?id=JxFgJbZ-wft
[ "Alexey Zakharov", "Qinghai Guo", "Zafeirios Fountas" ]
Poster
null
Discovery and learning of an underlying spatiotemporal hierarchy in sequential data is an important topic for machine learning. Despite this, little work has been done to explore hierarchical generative models that can flexibly adapt their layerwise representations in response to datasets with different temporal dynami...
[ "Hierarchical temporal abstraction", "event discovery", "hierarchical generative models", "variational inference" ]
null
4,647
2110.11236
title_snapshot
[ -0.01269999798387289, -0.009414407424628735, -0.0018756991485133767, 0.03752613440155983, 0.02997247688472271, 0.024535518139600754, 0.03070507012307644, 0.01948447711765766, -0.06229367479681969, -0.050385378301143646, 0.016227563843131065, -0.025907596573233604, -0.03206397220492363, 0.0...
Sample and Computation Redistribution for Efficient Face Detection
https://openreview.net/forum?id=RhB1AdoFfGE
[ "Jia Guo", "Jiankang Deng", "Alexandros Lattas", "Stefanos Zafeiriou" ]
Poster
null
Although tremendous strides have been made in uncontrolled face detection, accurate face detection with a low computation cost remains an open challenge. In this paper, we point out that computation distribution and scale augmentation are the keys to detecting small faces from low-resolution images. Motivated by these ...
[ "efficient face detection", "computation redistribution", "sample redistribution" ]
null
4,630
2105.04714
title_snapshot
[ -0.016903672367334366, -0.010471563786268234, 0.022874897345900536, 0.04426096752285957, 0.01434419583529234, 0.041566815227270126, 0.009826885536313057, -0.00738453259691596, -0.0075196437537670135, -0.06361646950244904, 0.002650728216394782, -0.010765195824205875, -0.06903128325939178, -...
Sound Adversarial Audio-Visual Navigation
https://openreview.net/forum?id=NkZq4OEYN-
[ "Yinfeng Yu", "Wenbing Huang", "Fuchun Sun", "Changan Chen", "Yikai Wang", "Xiaohong Liu" ]
Poster
null
Audio-visual navigation task requires an agent to find a sound source in a realistic, unmapped 3D environment by utilizing egocentric audio-visual observations. Existing audio-visual navigation works assume a clean environment that solely contains the target sound, which, however, would not be suitable in most real-wor...
[]
null
4,629
2202.10910
title_snapshot
[ -0.004416897892951965, 0.01641908660531044, 0.007598303724080324, 0.04555441439151764, 0.018972620368003845, 0.026937277987599373, 0.041419461369514465, 0.0015065225306898355, -0.05980093032121658, -0.0708814337849617, -0.019294507801532745, 0.028665946796536446, -0.05830734223127365, 0.00...
Out-of-distribution Generalization in the Presence of Nuisance-Induced Spurious Correlations
https://openreview.net/forum?id=12RoR2o32T
[ "Aahlad Manas Puli", "Lily H Zhang", "Eric Karl Oermann", "Rajesh Ranganath" ]
Poster
null
In many prediction problems, spurious correlations are induced by a changing relationship between the label and a nuisance variable that is also correlated with the covariates. For example, in classifying animals in natural images, the background, which is a nuisance, can predict the type of animal. This nuisance-label...
[ "spurious correlations", "out of distribution generalization", "ml for health", "representation learning" ]
null
4,618
2107.00520
title_snapshot
[ 0.00418870709836483, -0.010123956948518753, -0.022947870194911957, 0.031505364924669266, 0.03640491142868996, -0.0037574884481728077, 0.024712620303034782, -0.006317146588116884, -0.035393357276916504, -0.024500839412212372, -0.027831586077809334, 0.0023346177767962217, -0.08632460236549377,...
Dynamics-Aware Comparison of Learned Reward Functions
https://openreview.net/forum?id=CALFyKVs87
[ "Blake Wulfe", "Logan Michael Ellis", "Jean Mercat", "Rowan Thomas McAllister", "Adrien Gaidon" ]
Spotlight
null
The ability to learn reward functions plays an important role in enabling the deployment of intelligent agents in the real world. However, $\textit{comparing}$ reward functions, for example as a means of evaluating reward learning methods, presents a challenge. Reward functions are typically compared by considering the...
[ "Reward Learning", "Inverse Reinforcement Learning", "Reinforcement Learning", "Comparing Reward Functions" ]
null
4,616
2201.10081
title_snapshot
[ -0.05324899032711983, -0.029763156548142433, -0.016801027581095695, 0.029309263452887535, 0.05127202346920967, 0.009127838537096977, 0.009140978567302227, -0.02273988164961338, -0.05631231144070625, -0.028035679832100868, -0.011870114132761955, 0.04806471988558769, -0.041961655020713806, -...
AEVA: Black-box Backdoor Detection Using Adversarial Extreme Value Analysis
https://openreview.net/forum?id=OM_lYiHXiCL
[ "Junfeng Guo", "Ang Li", "Cong Liu" ]
Poster
null
Deep neural networks (DNNs) are proved to be vulnerable against backdoor attacks. A backdoor could be embedded in the target DNNs through injecting a backdoor trigger into the training examples, which can cause the target DNNs misclassify an input attached with the backdoor trigger. Recent backdoor detection methods o...
[]
null
4,615
2110.14880
title_snapshot
[ -0.0056268577463924885, -0.005832336843013763, -0.026485949754714966, 0.038314275443553925, 0.036808330565690994, 0.026126693934202194, 0.03938126936554909, -0.02523672953248024, -0.0171570535749197, -0.02851373329758644, -0.014844976365566254, 0.0062307012267410755, -0.03662409260869026, ...
Resonance in Weight Space: Covariate Shift Can Drive Divergence of SGD with Momentum
https://openreview.net/forum?id=5ECQL05ub0J
[ "Kirby Banman", "Garnet Liam Peet-Pare", "Nidhi Hegde", "Alona Fyshe", "Martha White" ]
Poster
null
Most convergence guarantees for stochastic gradient descent with momentum (SGDm) rely on iid sampling. Yet, SGDm is often used outside this regime, in settings with temporally correlated input samples such as continual learning and reinforcement learning. Existing work has shown that SGDm with a decaying step-size can...
[ "optimization", "momentum", "stochastic gradient descent", "non-iid sampling" ]
null
4,609
2203.11992
title_snapshot
[ -0.05744333192706108, -0.02202136628329754, 0.004181667231023312, 0.03815341740846634, 0.043017517775297165, 0.029292693361639977, 0.041487131267786026, -0.005730779841542244, -0.016667356714606285, -0.05203985050320625, 0.026844676584005356, -0.011875849217176437, -0.04458818957209587, 0....
Domino: Discovering Systematic Errors with Cross-Modal Embeddings
https://openreview.net/forum?id=FPCMqjI0jXN
[ "Sabri Eyuboglu", "Maya Varma", "Khaled Kamal Saab", "Jean-Benoit Delbrouck", "Christopher Lee-Messer", "Jared Dunnmon", "James Zou", "Christopher Re" ]
Oral
null
Machine learning models that achieve high overall accuracy often make systematic errors on important subsets (or slices) of data. Identifying underperforming slices is particularly challenging when working with high-dimensional inputs (e.g. images, audio), where important slices are often unlabeled. In order to address...
[ "robustness", "subgroup analysis", "error analysis", "multimodal", "slice discovery" ]
null
4,597
2203.14960
title_snapshot
[ -0.013684914447367191, -0.03830214962363243, -0.03655736893415451, 0.03311461955308914, 0.03555211052298546, 0.02598305605351925, 0.04597466066479683, 0.007174837402999401, -0.014069600962102413, -0.03622017055749893, -0.015240659937262535, 0.03504875302314758, -0.04507268965244293, -0.004...
Top-label calibration and multiclass-to-binary reductions
https://openreview.net/forum?id=WqoBaaPHS-
[ "Chirag Gupta", "Aaditya Ramdas" ]
Poster
null
We propose a new notion of multiclass calibration called top-label calibration. A classifier is said to be top-label calibrated if the reported probability for the predicted class label---the top-label---is calibrated, conditioned on the top-label. This conditioning is essential for practical utility of the calibration...
[ "calibration", "multiclass", "uncertainty quantification", "distribution-free", "histogram binning" ]
null
4,592
2107.08353
title_snapshot
[ 0.004305865149945021, 0.011524045839905739, -0.0030952105298638344, 0.037263330072164536, 0.0246293768286705, 0.013101226650178432, 0.004950529430061579, -0.018610630184412003, -0.02312423288822174, -0.04519007354974747, -0.02734309434890747, -0.007206159643828869, -0.06907548755407333, 0....
Anisotropic Random Feature Regression in High Dimensions
https://openreview.net/forum?id=JfaWawZ8BmX
[ "Gabriel Mel", "Jeffrey Pennington" ]
Poster
null
In contrast to standard statistical wisdom, modern learning algorithms typically find their best performance in the overparameterized regime in which the model has many more parameters than needed to fit the training data. A growing number of recent works have shown that random feature models can offer a detailed theor...
[ "random feature models", "high dimensional asymptotics", "generalization", "learning curves", "double descent", "multiple descent", "alignment" ]
null
4,589
null
null
[ -0.008536995388567448, -0.020025094971060753, 0.009098321199417114, 0.035476382821798325, 0.01325021218508482, 0.030345261096954346, 0.034612152725458145, -0.01824302040040493, -0.007867624051868916, -0.061600759625434875, -0.015139257535338402, 0.0016697929240763187, -0.05730269476771355, ...
Back2Future: Leveraging Backfill Dynamics for Improving Real-time Predictions in Future
https://openreview.net/forum?id=L01Nn_VJ9i
[ "Harshavardhan Kamarthi", "Alexander Rodríguez", "B. Aditya Prakash" ]
Poster
null
For real-time forecasting in domains like public health and macroeconomics, data collection is a non-trivial and demanding task. Often after being initially released, it undergoes several revisions later (maybe due to human or technical constraints) - as a result, it may take weeks until the data reaches a stable value...
[ "Epidemic Forecasting", "Data revisions", "Graph Representation learning", "Time Series Forecasting" ]
null
4,586
2106.04420
title_snapshot
[ -0.010129773989319801, -0.05491367354989052, -0.004140959586948156, -0.007232226897031069, 0.0386655330657959, 0.03353913500905037, 0.009027084335684776, 0.051223836839199066, -0.033514153212308884, -0.04910697042942047, 0.010371112264692783, -0.006898882798850536, -0.04648352414369583, 0....
Approximation and Learning with Deep Convolutional Models: a Kernel Perspective
https://openreview.net/forum?id=lrocYB-0ST2
[ "Alberto Bietti" ]
Poster
null
The empirical success of deep convolutional networks on tasks involving high-dimensional data such as images or audio suggests that they can efficiently approximate certain functions that are well-suited for such tasks. In this paper, we study this through the lens of kernel methods, by considering simple hierarchical ...
[ "kernel methods", "deep learning theory", "convolution", "approximation", "generalization" ]
null
4,570
2102.10032
title_snapshot
[ 0.010854906402528286, -0.026529820635914803, 0.029386000707745552, 0.0560007318854332, 0.024497414007782936, 0.049019478261470795, 0.016565969213843346, -0.008924144320189953, -0.012564441189169884, -0.029380429536104202, -0.01201192568987608, 0.0068969521671533585, -0.056928034871816635, ...
Value Function Spaces: Skill-Centric State Abstractions for Long-Horizon Reasoning
https://openreview.net/forum?id=vgqS1vkkCbE
[ "Dhruv Shah", "Peng Xu", "Yao Lu", "Ted Xiao", "Alexander T Toshev", "Sergey Levine", "brian ichter" ]
Poster
null
Reinforcement learning can train policies that effectively perform complex tasks. However for long-horizon tasks, the performance of these methods degrades with horizon, often necessitating reasoning over and chaining lower-level skills. Hierarchical reinforcement learning aims to enable this by providing a bank of low...
[ "hierarchical reinforcement learning", "planning", "representation learning", "robotics" ]
null
4,569
2111.03189
title_snapshot
[ -0.04057324305176735, 0.004519219975918531, -0.011432568542659283, 0.04205481335520744, 0.06761135905981064, 0.029569149017333984, 0.011927658692002296, -0.015531035140156746, -0.036929622292518616, -0.034943509846925735, -0.022663990035653114, 0.024249570444226265, -0.07815144211053848, -...
Natural Language Descriptions of Deep Visual Features
https://openreview.net/forum?id=NudBMY-tzDr
[ "Evan Hernandez", "Sarah Schwettmann", "David Bau", "Teona Bagashvili", "Antonio Torralba", "Jacob Andreas" ]
Oral
null
Some neurons in deep networks specialize in recognizing highly specific perceptual, structural, or semantic features of inputs. In computer vision, techniques exist for identifying neurons that respond to individual concept categories like colors, textures, and object classes. But these techniques are limited in scope,...
[]
null
4,559
2201.11114
title_snapshot
[ -0.030006742104887962, -0.0027836693916469812, -0.024491438642144203, 0.05133626610040665, 0.03981329873204231, 0.03815820440649986, 0.012636391445994377, 0.03732728213071823, -0.029389822855591774, -0.02512463927268982, -0.07315713167190552, 0.003692387603223324, -0.040563713759183884, -0...
Learning Hierarchical Structures with Differentiable Nondeterministic Stacks
https://openreview.net/forum?id=5LXw_QplBiF
[ "Brian DuSell", "David Chiang" ]
Spotlight
null
Learning hierarchical structures in sequential data -- from simple algorithmic patterns to natural language -- in a reliable, generalizable way remains a challenging problem for neural language models. Past work has shown that recurrent neural networks (RNNs) struggle to generalize on held-out algorithmic or syntactic ...
[ "RNN", "pushdown automata", "nondeterminism", "formal languages", "language modeling" ]
null
4,554
2109.01982
title_snapshot
[ -0.0434621162712574, -0.0005775233148597181, -0.018160581588745117, 0.04565917328000069, 0.02690124325454235, 0.05930456146597862, 0.037486981600522995, 0.015870744362473488, -0.05532574653625488, -0.016881423071026802, 0.013759495690464973, -0.00006037799903424457, -0.054962094873189926, ...
Fast Regression for Structured Inputs
https://openreview.net/forum?id=gNp54NxHUPJ
[ "Raphael A Meyer", "Cameron N Musco", "Christopher P Musco", "David Woodruff", "Samson Zhou" ]
Poster
null
We study the $\ell_p$ regression problem, which requires finding $\mathbf{x}\in\mathbb R^{d}$ that minimizes $\|\mathbf{A}\mathbf{x}-\mathbf{b}\|_p$ for a matrix $\mathbf{A}\in\mathbb R^{n \times d}$ and response vector $\mathbf{b}\in\mathbb R^{n}$. There has been recent interest in developing subsampling methods for t...
[ "regression", "sublinear time algorithm", "structured input" ]
null
4,544
2203.07557
title_snapshot
[ -0.012141361832618713, -0.019982576370239258, 0.0021919880528002977, 0.0210301224142313, 0.047066863626241684, 0.03729786351323128, 0.022094611078500748, -0.024738462641835213, -0.005064369644969702, -0.027681611478328705, -0.0160700511187315, 0.009534159675240517, -0.08239665627479553, -0...
CrossBeam: Learning to Search in Bottom-Up Program Synthesis
https://openreview.net/forum?id=qhC8mr2LEKq
[ "Kensen Shi", "Hanjun Dai", "Kevin Ellis", "Charles Sutton" ]
Poster
null
Many approaches to program synthesis perform a search within an enormous space of programs to find one that satisfies a given specification. Prior works have used neural models to guide combinatorial search algorithms, but such approaches still explore a huge portion of the search space and quickly become intractable a...
[ "Program Synthesis", "Bottom-Up Search" ]
null
4,543
2203.10452
title_snapshot
[ -0.031020034104585648, -0.00924767553806305, -0.031914807856082916, 0.015852639451622963, 0.03752977028489113, 0.026991404592990875, 0.0089692622423172, 0.0022599874064326286, -0.030018581077456474, -0.030736975371837616, -0.0026566176675260067, 0.02316165156662464, -0.04703168943524361, -...
PEARL: Data Synthesis via Private Embeddings and Adversarial Reconstruction Learning
https://openreview.net/forum?id=M6M8BEmd6dq
[ "Seng Pei Liew", "Tsubasa Takahashi", "Michihiko Ueno" ]
Poster
null
We propose a new framework of synthesizing data using deep generative models in a differentially private manner. Within our framework, sensitive data are sanitized with rigorous privacy guarantees in a one-shot fashion, such that training deep generative models is possible without re-using the original data. Hence, no ...
[ "Differential Privacy", "Generative Model" ]
null
4,542
2106.04590
title_snapshot
[ 0.006481783464550972, -0.018035385757684708, -0.02104363963007927, 0.08799503743648529, 0.0664215236902237, 0.04572538286447525, 0.04189899563789368, -0.03407259285449982, -0.004177049733698368, -0.013612891547381878, -0.0034402792807668447, 0.005394540261477232, -0.07153838872909546, -0.0...
Divisive Feature Normalization Improves Image Recognition Performance in AlexNet
https://openreview.net/forum?id=aOX3a9q3RVV
[ "Michelle Miller", "SueYeon Chung", "Kenneth D. Miller" ]
Poster
null
Local divisive normalization provides a phenomenological description of many nonlinear response properties of neurons across visual cortical areas. To gain insight into the utility of this operation, we studied the effects on AlexNet of a local divisive normalization between features, with learned parameters. Developin...
[ "divisive normalization", "AlexNet", "ImageNet", "CIFAR-100", "manifold capacity", "sparsity", "receptive fields", "Batch Normalization", "Group Normalization", "Layer Normalization" ]
null
4,523
null
null
[ 0.009878062643110752, -0.0020991351921111345, -0.0049480558373034, 0.01469709724187851, 0.024717899039387703, 0.023739023134112358, 0.02641955390572548, -0.0003358266840223223, -0.04386884719133377, -0.05026276037096977, 0.007348506711423397, -0.03776496648788452, -0.07199331372976303, -0....
Evaluating Distributional Distortion in Neural Language Modeling
https://openreview.net/forum?id=bTteFbU99ye
[ "Benjamin LeBrun", "Alessandro Sordoni", "Timothy J. O'Donnell" ]
Poster
null
A fundamental characteristic of natural language is the high rate at which speakers produce novel expressions. Because of this novelty, a heavy-tail of rare events accounts for a significant amount of the total probability mass of distributions in language (Baayen, 2001). Standard language modeling metrics such as perp...
[]
null
4,509
2203.12788
title_snapshot
[ -0.013415650464594364, 0.0024214067962020636, -0.03107968345284462, 0.041972897946834564, 0.041502222418785095, 0.04359380528330803, 0.028992416337132454, 0.030400915071368217, -0.01570982113480568, 0.00042438795207999647, -0.021742718294262886, 0.04040141776204109, -0.03843219205737114, -...
MaGNET: Uniform Sampling from Deep Generative Network Manifolds Without Retraining
https://openreview.net/forum?id=r5qumLiYwf9
[ "Ahmed Imtiaz Humayun", "Randall Balestriero", "Richard Baraniuk" ]
Poster
null
Deep Generative Networks (DGNs) are extensively employed in Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and their variants to approximate the data manifold, and data distribution on that manifold. However, training samples are often obtained based on preferences, costs, or convenience produ...
[ "Deep Generative Networks", "Uniform Sampling", "Fairness", "Data Augmentation" ]
null
4,501
2110.08009
title_snapshot
[ -0.010706021450459957, -0.022258389741182327, 0.0013295183889567852, 0.06948699802160263, 0.005251140333712101, 0.026060957461595535, 0.02506859041750431, -0.002526527503505349, -0.017215268686413765, -0.05228189751505852, -0.009161386638879776, -0.02020731195807457, -0.09348389506340027, ...
Sampling with Mirrored Stein Operators
https://openreview.net/forum?id=eMudnJsb1T5
[ "Jiaxin Shi", "Chang Liu", "Lester Mackey" ]
Spotlight
null
We introduce a new family of particle evolution samplers suitable for constrained domains and non-Euclidean geometries. Stein Variational Mirror Descent and Mirrored Stein Variational Gradient Descent minimize the Kullback-Leibler (KL) divergence to constrained target distributions by evolving particles in a dual space...
[ "Stein's method", "Sampling", "Mirror descent", "Natural gradient descent", "Probabilistic inference", "Bayesian inference", "Post-selection inference", "Stein operators" ]
null
4,500
2106.12506
title_snapshot
[ -0.014372757636010647, -0.012093458324670792, 0.03477240726351738, 0.025930996984243393, 0.014311743900179863, 0.01904216594994068, 0.02289826236665249, 0.005761237815022469, -0.02980811707675457, -0.05960613489151001, -0.008860204368829727, -0.01767011359333992, -0.056256331503391266, 0.0...
Planning in Stochastic Environments with a Learned Model
https://openreview.net/forum?id=X6D9bAHhBQ1
[ "Ioannis Antonoglou", "Julian Schrittwieser", "Sherjil Ozair", "Thomas K Hubert", "David Silver" ]
Spotlight
null
Model-based reinforcement learning has proven highly successful. However, learning a model in isolation from its use during planning is problematic in complex environments. To date, the most effective techniques have instead combined value-equivalent model learning with powerful tree-search methods. This approach is ex...
[ "model-based reinforcement learning", "deep reinforcement learning", "tree based search", "MCTS" ]
null
4,498
null
null
[ -0.022786185145378113, -0.018640922382473946, -0.007135044317692518, 0.03125857561826706, 0.05290883406996727, 0.020049525424838066, 0.021024199202656746, 0.01381091307848692, -0.043486882001161575, -0.05375121533870697, 0.008105874061584473, 0.005191822070628405, -0.073745496571064, -0.01...
Neural Contextual Bandits with Deep Representation and Shallow Exploration
https://openreview.net/forum?id=xnYACQquaGV
[ "Pan Xu", "Zheng Wen", "Handong Zhao", "Quanquan Gu" ]
Poster
null
We study neural contextual bandits, a general class of contextual bandits, where each context-action pair is associated with a raw feature vector, but the specific reward generating function is unknown. We propose a novel learning algorithm that transforms the raw feature vector using the last hidden layer of a deep Re...
[ "neural network", "deep representation learning" ]
null
4,495
2012.01780
title_snapshot
[ -0.03030621074140072, -0.011163175106048584, -0.012585600838065147, 0.05720676854252815, 0.02102486602962017, 0.030277356505393982, 0.030278554186224937, 0.021160440519452095, -0.03802670165896416, -0.03130415081977844, -0.030390571802854538, 0.008680158294737339, -0.03677933290600777, -0....
PI3NN: Out-of-distribution-aware Prediction Intervals from Three Neural Networks
https://openreview.net/forum?id=NoB8YgRuoFU
[ "Siyan Liu", "Pei Zhang", "Dan Lu", "Guannan Zhang" ]
Poster
null
We propose a novel prediction interval (PI) method for uncertainty quantification, which addresses three major issues with the state-of-the-art PI methods. First, existing PI methods require retraining of neural networks (NNs) for every given confidence level and suffer from the crossing issue in calculating multiple P...
[]
null
4,494
2108.02327
title_snapshot
[ -0.010704757645726204, 0.008360964246094227, -0.03439963236451149, 0.03988145664334297, 0.04283597320318222, 0.02800147235393524, 0.009173436090350151, 0.005549705121666193, -0.03341837599873543, -0.04407966509461403, 0.03240760788321495, 0.000022218417143449187, -0.07886768877506256, -0.0...
Non-Transferable Learning: A New Approach for Model Ownership Verification and Applicability Authorization
https://openreview.net/forum?id=tYRrOdSnVUy
[ "Lixu Wang", "Shichao Xu", "Ruiqi Xu", "Xiao Wang", "Qi Zhu" ]
Oral
null
As Artificial Intelligence as a Service gains popularity, protecting well-trained models as intellectual property is becoming increasingly important. There are two common types of protection methods: ownership verification and usage authorization. In this paper, we propose Non-Transferable Learning (NTL), a novel appro...
[ "Domain Adaptation", "Transfer Learning", "Societal Considerations of Representation Learning", "Model Watermark" ]
null
4,491
2106.06916
title_snapshot
[ 0.008621430955827236, -0.02197086624801159, -0.021497618407011032, 0.039983950555324554, 0.06121435761451721, -0.00898057036101818, 0.008286764845252037, -0.013322819024324417, -0.006241966504603624, -0.013649032451212406, -0.017900919541716576, 0.001923100557178259, -0.04832291975617409, ...
Discriminative Similarity for Data Clustering
https://openreview.net/forum?id=kj0_45Y4r9i
[ "Yingzhen Yang", "Ping Li" ]
Poster
null
Similarity-based clustering methods separate data into clusters according to the pairwise similarity between the data, and the pairwise similarity is crucial for their performance. In this paper, we propose {\em Clustering by Discriminative Similarity (CDS)}, a novel method which learns discriminative similarity for d...
[ "Discriminative Similarity", "Rademacher Complexity", "Generalization Bound", "Data Clustering" ]
null
4,489
2109.08675
title_snapshot
[ -0.018822064623236656, -0.029979560524225235, 0.011519442312419415, 0.058707188814878464, 0.05003536120057106, 0.019266873598098755, -0.012640438042581081, -0.02441941760480404, 0.01202161144465208, -0.04002854600548744, -0.03322356194257736, 0.007362199481576681, -0.0782868042588234, 0.04...
It Takes Four to Tango: Multiagent Self Play for Automatic Curriculum Generation
https://openreview.net/forum?id=q4tZR1Y-UIs
[ "Yuqing Du", "Pieter Abbeel", "Aditya Grover" ]
Poster
null
We are interested in training general-purpose reinforcement learning agents that can solve a wide variety of goals. Training such agents efficiently requires automatic generation of a goal curriculum. This is challenging as it requires (a) exploring goals of increasing difficulty, while ensuring that the agent (b) is e...
[ "curriculum generation", "unsupervised reinforcement learning", "goal conditioned reinforcement learning", "multi agent" ]
null
4,485
2202.10608
title_judge
[ -0.015511061996221542, -0.03532794117927551, -0.0099917221814394, 0.04387117922306061, 0.04553936794400215, 0.003257278585806489, 0.036399245262145996, 0.00011871907918248326, -0.023819567635655403, -0.025706423446536064, -0.02375224605202675, 0.033189550042152405, -0.055922918021678925, -...
CROP: Certifying Robust Policies for Reinforcement Learning through Functional Smoothing
https://openreview.net/forum?id=HOjLHrlZhmx
[ "Fan Wu", "Linyi Li", "Zijian Huang", "Yevgeniy Vorobeychik", "Ding Zhao", "Bo Li" ]
Poster
null
As reinforcement learning (RL) has achieved great success and been even adopted in safety-critical domains such as autonomous vehicles, a range of empirical studies have been conducted to improve its robustness against adversarial attacks. However, how to certify its robustness with theoretical guarantees still remains...
[]
null
4,469
2106.09292
title_snapshot
[ -0.01116949412971735, -0.03533653914928436, -0.010070697404444218, 0.043783530592918396, 0.03570641949772835, 0.02137545309960842, 0.0049577392637729645, -0.015815937891602516, -0.018885036930441856, -0.024889253079891205, -0.02601383626461029, 0.010728362016379833, -0.07071583718061447, -...
Neural Link Prediction with Walk Pooling
https://openreview.net/forum?id=CCu6RcUMwK0
[ "Liming Pan", "Cheng Shi", "Ivan Dokmanić" ]
Poster
null
Graph neural networks achieve high accuracy in link prediction by jointly leveraging graph topology and node attributes. Topology, however, is represented indirectly; state-of-the-art methods based on subgraph classification label nodes with distance to the target link, so that, although topological information is pres...
[ "Graph neural network", "Link prediction", "Random walk", "Graph topology." ]
null
4,448
2110.04375
title_snapshot
[ 0.013805821537971497, -0.046472061425447464, -0.00119615881703794, 0.030126696452498436, 0.028501268476247787, 0.03027448244392872, 0.0071074990555644035, 0.024677427485585213, -0.010724742896854877, -0.057407528162002563, 0.005734946578741074, -0.02999592013657093, -0.053851451724767685, ...
On the Convergence of Certified Robust Training with Interval Bound Propagation
https://openreview.net/forum?id=YeShU5mLfLt
[ "Yihan Wang", "Zhouxing Shi", "Quanquan Gu", "Cho-Jui Hsieh" ]
Poster
null
Interval Bound Propagation (IBP) is so far the base of state-of-the-art methods for training neural networks with certifiable robustness guarantees when potential adversarial perturbations present, while the convergence of IBP training remains unknown in existing literature. In this paper, we present a theoretical anal...
[ "Certified robustness", "Adversarial robustness", "Convergence" ]
null
4,436
2203.08961
title_snapshot
[ -0.033936064690351486, -0.00785810686647892, -0.014392203651368618, 0.027603816241025925, 0.03301288187503815, 0.015080887824296951, 0.050762053579092026, -0.009424780495464802, -0.041075821965932846, -0.011020757257938385, -0.0036414233036339283, 0.01872473955154419, -0.07434579730033875, ...
Pretraining Text Encoders with Adversarial Mixture of Training Signal Generators
https://openreview.net/forum?id=sX3XaHwotOg
[ "Yu Meng", "Chenyan Xiong", "Payal Bajaj", "saurabh tiwary", "Paul N. Bennett", "Jiawei Han", "Xia Song" ]
Poster
null
We present a new framework AMOS that pretrains text encoders with an Adversarial learning curriculum via a Mixture Of Signals from multiple auxiliary generators. Following ELECTRA-style pretraining, the main encoder is trained as a discriminator to detect replaced tokens generated by auxiliary masked language models (M...
[ "Language Model Pretraining" ]
null
4,429
2204.03243
title_snapshot
[ -0.0009753016056492925, -0.04366389662027359, -0.014068893156945705, 0.07657379657030106, 0.020338138565421104, 0.02323155663907528, 0.024740636348724365, -0.0040457420982420444, -0.040474921464920044, -0.013143850490450859, -0.058377474546432495, 0.049213480204343796, -0.0520717054605484, ...
Neural Structured Prediction for Inductive Node Classification
https://openreview.net/forum?id=YWNAX0caEjI
[ "Meng Qu", "Huiyu Cai", "Jian Tang" ]
Oral
null
This paper studies node classification in the inductive setting, i.e., aiming to learn a model on labeled training graphs and generalize it to infer node labels on unlabeled test graphs. This problem has been extensively studied with graph neural networks (GNNs) by learning effective node representations, as well as tr...
[]
null
4,412
2204.07524
title_snapshot
[ -0.0018082709284499288, -0.04475924000144005, 0.010683368891477585, 0.04850548133254051, 0.03372834622859955, 0.02824607491493225, -0.003414466977119446, -0.02211669273674488, -0.021362993866205215, -0.01935308799147606, 0.034616973251104355, 0.0028129820711910725, -0.0699390321969986, 0.0...
Towards Training Billion Parameter Graph Neural Networks for Atomic Simulations
https://openreview.net/forum?id=0jP2n0YFmKG
[ "Anuroop Sriram", "Abhishek Das", "Brandon M Wood", "Siddharth Goyal", "C. Lawrence Zitnick" ]
Poster
null
Recent progress in Graph Neural Networks (GNNs) for modeling atomic simulations has the potential to revolutionize catalyst discovery, which is a key step in making progress towards the energy breakthroughs needed to combat climate change. However, the GNNs that have proven most effective for this task are memory inten...
[ "Graph Neural Networks", "Atomic Simulations", "Computational Chemistry" ]
null
4,408
2203.09697
title_snapshot
[ -0.025483524426817894, -0.006781612057238817, -0.009506180882453918, 0.03740974888205528, 0.019006896764039993, 0.02684224396944046, 0.013529259711503983, 0.005562251899391413, -0.010162625461816788, -0.03974059224128723, 0.04087331146001816, -0.03034214675426483, -0.07346558570861816, 0.0...
RotoGrad: Gradient Homogenization in Multitask Learning
https://openreview.net/forum?id=T8wHz4rnuGL
[ "Adrián Javaloy", "Isabel Valera" ]
Spotlight
null
Multitask learning is being increasingly adopted in applications domains like computer vision and reinforcement learning. However, optimally exploiting its advantages remains a major challenge due to the effect of negative transfer. Previous works have tracked down this issue to the disparities in gradient magnitudes a...
[ "multitask learning", "conflicting gradients", "negative transfer" ]
null
4,402
2103.02631
title_snapshot
[ 0.012622673995792866, -0.02841230481863022, 0.006820524577051401, 0.030286842957139015, 0.014274388551712036, 0.04120655730366707, -0.0034386287443339825, -0.02101299725472927, -0.01790446788072586, -0.06566285341978073, 0.007264528889209032, 0.0031944813672453165, -0.0615457147359848, -0....
On Improving Adversarial Transferability of Vision Transformers
https://openreview.net/forum?id=D6nH3719vZy
[ "Muzammal Naseer", "Kanchana Ranasinghe", "Salman Khan", "Fahad Khan", "Fatih Porikli" ]
Spotlight
null
Vision transformers (ViTs) process input images as sequences of patches via self-attention; a radically different architecture than convolutional neural networks (CNNs). This makes it interesting to study the adversarial feature space of ViT models and their transferability. In particular, we observe that adversarial ...
[ "Vision Transformers", "Adversarial Perturbations" ]
null
4,395
2106.04169
title_snapshot
[ 0.0066090417094528675, -0.02235463075339794, 0.0015845258021727204, 0.025110797956585884, 0.02014222927391529, 0.03179061412811279, 0.023536505177617073, 0.00030260166386142373, -0.0004583622794598341, -0.03192513436079025, -0.03133275732398033, 0.026754993945360184, -0.06424453109502792, ...
On Predicting Generalization using GANs
https://openreview.net/forum?id=eW5R4Cek6y6
[ "Yi Zhang", "Arushi Gupta", "Nikunj Saunshi", "Sanjeev Arora" ]
Spotlight
null
Research on generalization bounds for deep networks seeks to give ways to predict test error using just the training dataset and the network parameters. While generalization bounds can give many insights about architecture design, training algorithms etc., what they do not currently do is yield good predictions for act...
[ "generalization", "generative adversarial network" ]
null
4,393
2111.14212
title_snapshot
[ -0.008986820466816425, -0.016538942232728004, -0.020124714821577072, 0.054934196174144745, 0.032410964369773865, 0.009826501831412315, 0.04132791608572006, -0.00317170238122344, -0.0013533049495890737, -0.0428926944732666, -0.007338551804423332, 0.003130767261609435, -0.07457426935434341, ...
Understanding and Leveraging Overparameterization in Recursive Value Estimation
https://openreview.net/forum?id=shbAgEsk3qM
[ "Chenjun Xiao", "Bo Dai", "Jincheng Mei", "Oscar A Ramirez", "Ramki Gummadi", "Chris Harris", "Dale Schuurmans" ]
Poster
null
The theory of function approximation in reinforcement learning (RL) typically considers low capacity representations that incur a tradeoff between approximation error, stability and generalization. Current deep architectures, however, operate in an overparameterized regime where approximation error is not necessarily ...
[ "Temporal Difference Learning", "Residual Minimization", "Value Estimation", "Overparameterization" ]
null
4,376
null
null
[ -0.02479463629424572, -0.040777821093797684, 0.005239318124949932, 0.04001184180378914, 0.051100954413414, 0.049285680055618286, 0.02926049940288067, -0.009250788018107414, -0.016967248171567917, -0.04107934236526489, -0.015228861011564732, 0.004978512413799763, -0.08651445806026459, -0.01...
Optimization and Adaptive Generalization of Three layer Neural Networks
https://openreview.net/forum?id=dPyRNUlttBv
[ "Khashayar Gatmiry", "Stefanie Jegelka", "Jonathan Kelner" ]
Poster
null
While there has been substantial recent work studying generalization of neural networks, the ability of deep nets in automating the process of feature extraction still evades a thorough mathematical understanding. As a step toward this goal, we analyze learning and generalization of a three-layer neural network wit...
[ "deep learning theory", "adaptive kernel", "robust deep learning", "neural tangent kernel", "adaptive generalization", "non-convex optimization" ]
null
4,368
null
null
[ -0.049622777849435806, -0.005750476848334074, 0.017237218096852303, 0.04321613535284996, 0.03622562438249588, 0.04609695449471474, 0.018869658932089806, 0.002591748023405671, -0.04181499034166336, -0.03192451596260071, -0.022727882489562035, 0.012251501902937889, -0.060234397649765015, 0.0...
Non-Parallel Text Style Transfer with Self-Parallel Supervision
https://openreview.net/forum?id=-TSe5o7STVR
[ "Ruibo Liu", "Chongyang Gao", "Chenyan Jia", "Guangxuan Xu", "Soroush Vosoughi" ]
Poster
null
The performance of existing text style transfer models is severely limited by the non-parallel datasets on which the models are trained. In non-parallel datasets, no direct mapping exists between sentences of the source and target style; the style transfer models thus only receive weak supervision of the target sentenc...
[ "style transfer", "non-parallel corpus", "imitation learning", "language models", "political stance transfer" ]
null
4,365
2204.08123
title_snapshot
[ 0.01724928244948387, -0.049092430621385574, -0.004230727441608906, 0.05097030848264694, 0.03332718834280968, 0.03755764290690422, -0.005295365583151579, 0.018585950136184692, 0.007795832119882107, -0.02342991530895233, -0.02572862058877945, 0.02022535726428032, -0.073183074593544, 0.019046...
Can an Image Classifier Suffice For Action Recognition?
https://openreview.net/forum?id=qhkFX-HLuHV
[ "Quanfu Fan", "Chun-Fu Chen", "Rameswar Panda" ]
Poster
null
We explore a new perspective on video understanding by casting the video recognition problem as an image recognition task. Our approach rearranges input video frames into super images, which allow for training an image classifier directly to fulfill the task of action recognition, in exactly the same way as image class...
[ "action recognition", "image classifier", "super image", "vision transformer" ]
null
4,364
2106.14104
title_snapshot
[ 0.023202946409583092, -0.03669794276356697, -0.01960579678416252, 0.028179701417684555, 0.01224860455840826, 0.023146413266658783, 0.029743976891040802, 0.007877880707383156, -0.019394895061850548, -0.02939269319176674, 0.009277617558836937, -0.007525424938648939, -0.05836789309978485, 0.0...
On the Connection between Local Attention and Dynamic Depth-wise Convolution
https://openreview.net/forum?id=L3_SsSNMmy
[ "Qi Han", "Zejia Fan", "Qi Dai", "Lei Sun", "Ming-Ming Cheng", "Jiaying Liu", "Jingdong Wang" ]
Spotlight
null
Vision Transformer (ViT) attains state-of-the-art performance in visual recognition, and the variant, Local Vision Transformer, makes further improvements. The major component in Local Vision Transformer, local attention, performs the attention separately over small local windows. We rephrase local attention as a chann...
[ "local attention", "depth-wise convolution", "dynamic depth-wise convolution", "weight sharing", "dynamic weight" ]
null
4,356
2106.04263
title_snapshot
[ -0.017382128164172173, -0.012365199625492096, 0.025403551757335663, 0.04119936376810074, 0.02960570529103279, 0.047041989862918854, 0.016303349286317825, 0.007533975876867771, -0.02316994033753872, -0.04449261724948883, -0.010240845382213593, 0.030654052272439003, -0.06986761838197708, 0.0...
Strength of Minibatch Noise in SGD
https://openreview.net/forum?id=uorVGbWV5sw
[ "Liu Ziyin", "Kangqiao Liu", "Takashi Mori", "Masahito Ueda" ]
Spotlight
null
The noise in stochastic gradient descent (SGD), caused by minibatch sampling, is poorly understood despite its practical importance in deep learning. This work presents the first systematic study of the SGD noise and fluctuations close to a local minimum. We first analyze the SGD noise in linear regression in detail an...
[ "stochastic gradient descent", "minibatch noise", "discrete-time SGD", "noise and fluctuation", "exact solvable models" ]
null
4,355
2102.05375
title_snapshot
[ -0.04124589264392853, -0.01990743912756443, -0.02109786681830883, 0.027885695919394493, 0.0395432710647583, 0.053466636687517166, 0.03313787281513214, 0.009385067969560623, -0.02297869697213173, -0.05617015063762665, 0.0009675398468971252, -0.01792357861995697, -0.05335870012640953, 0.0166...
Learning more skills through optimistic exploration
https://openreview.net/forum?id=cU8rknuhxc
[ "DJ Strouse", "Kate Baumli", "David Warde-Farley", "Volodymyr Mnih", "Steven Stenberg Hansen" ]
Spotlight
null
Unsupervised skill learning objectives (Eysenbach et al., 2019; Gregor et al., 2016) allow agents to learn rich repertoires of behavior in the absence of extrinsic rewards. They work by simultaneously training a policy to produce distinguishable latent-conditioned trajectories, and a discriminator to evaluate distingui...
[ "intrinsic control", "skill discovery", "unsupervised skill learning", "uncertainty estimation", "optimistic exploration", "variational information maximization" ]
null
4,352
2107.14226
title_snapshot
[ -0.026958152651786804, -0.01341957040131092, -0.008654285222291946, 0.06747882068157196, 0.026265576481819153, 0.014448538422584534, 0.03431682288646698, -0.008682113140821457, -0.03077227808535099, -0.040427546948194504, -0.02421596460044384, 0.019561640918254852, -0.06448059529066086, -0...
Interacting Contour Stochastic Gradient Langevin Dynamics
https://openreview.net/forum?id=IK9ap6nxXr2
[ "Wei Deng", "Siqi Liang", "Botao Hao", "Guang Lin", "Faming Liang" ]
Poster
null
We propose an interacting contour stochastic gradient Langevin dynamics (ICSGLD) sampler, an embarrassingly parallel multiple-chain contour stochastic gradient Langevin dynamics (CSGLD) sampler with efficient interactions. We show that ICSGLD can be theoretically more efficient than a single-chain CSGLD with an equival...
[ "stochastic gradient Langevin dynamics", "MCMC", "importance sampling", "Wang-Landau algorithm", "Parallel MCMC Methods", "stochastic approximation" ]
null
4,326
2202.09867
title_snapshot
[ 0.01764717325568199, 0.0028313565999269485, -0.0008361748768948019, 0.0625057965517044, 0.03601701185107231, 0.03568458929657936, 0.023550741374492645, 0.0125583466142416, -0.03214813396334648, -0.04852182790637016, 0.018526045605540276, -0.008318236097693443, -0.06620123982429504, -0.0141...
NeuPL: Neural Population Learning
https://openreview.net/forum?id=MIX3fJkl_1
[ "Siqi Liu", "Luke Marris", "Daniel Hennes", "Josh Merel", "Nicolas Heess", "Thore Graepel" ]
Poster
null
Learning in strategy games (e.g. StarCraft, poker) requires the discovery of diverse policies. This is often achieved by iteratively training new policies against existing ones, growing a policy population that is robust to exploit. This iterative approach suffers from two issues in real-world games: a) under finite bu...
[ "Multi-Agent Learning", "Game Theory", "Population Learning" ]
null
4,325
2202.07415
title_snapshot
[ -0.052040304988622665, -0.037821173667907715, 0.01761529967188835, 0.029755404219031334, 0.03604955971240997, 0.03906282037496567, -0.010761252604424953, -0.0007639609975740314, -0.04809200018644333, -0.03168457746505737, 0.002590855350717902, 0.015367339365184307, -0.06501863896846771, -0...
Reinforcement Learning under a Multi-agent Predictive State Representation Model: Method and Theory
https://openreview.net/forum?id=PLDOnFoVm4
[ "Zhi Zhang", "Zhuoran Yang", "Han Liu", "Pratap Tokekar", "Furong Huang" ]
Spotlight
null
We study reinforcement learning for partially observable multi-agent systems where each agent only has access to its own observation and reward and aims to maximize its cumulative rewards. To handle partial observations, we propose graph-assisted predictive state representations (GAPSR), a scalable multi-agent represen...
[ "Multi-agent Reinforcement Learning", "Predictive State Representation", "Dynamic Interaction Graph" ]
null
4,323
null
null
[ -0.03968605399131775, -0.013906953856348991, -0.004196410067379475, 0.056067366153001785, 0.04592905193567276, 0.018528863787651062, 0.02025405876338482, 0.01751355081796646, -0.04391845315694809, -0.02557811141014099, 0.006738792639225721, -0.00319076981395483, -0.09156890213489532, -0.00...
A New Perspective on "How Graph Neural Networks Go Beyond Weisfeiler-Lehman?"
https://openreview.net/forum?id=uxgg9o7bI_3
[ "Asiri Wijesinghe", "Qing Wang" ]
Oral
null
We propose a new perspective on designing powerful Graph Neural Networks (GNNs). In a nutshell, this enables a general solution to inject structural properties of graphs into a message-passing aggregation scheme of GNNs. As a theoretical basis, we develop a new hierarchy of local isomorphism on neighborhood subgraphs. ...
[ "Graph Neural Networks", "Graph Isomorphism", "Weisfeiler Lehman" ]
null
4,320
null
null
[ -0.026146888732910156, -0.014835226349532604, 0.022459980100393295, 0.053130246698856354, 0.03560756519436836, 0.019757740199565887, 0.022399039939045906, 0.020890498533844948, -0.004813455976545811, -0.02768111415207386, 0.038211412727832794, -0.010243845172226429, -0.0719047263264656, 0....
DeSKO: Stability-Assured Robust Control with a Deep Stochastic Koopman Operator
https://openreview.net/forum?id=hniLRD_XCA
[ "Minghao Han", "Jacob Euler-Rolle", "Robert K. Katzschmann" ]
Poster
null
The Koopman operator theory linearly describes nonlinear dynamical systems in a high-dimensional functional space and it allows to apply linear control methods to highly nonlinear systems. However, the Koopman operator does not account for any uncertainty in dynamical systems, causing it to perform poorly in real-world...
[ "Koopman Operator", "Robust Control", "Robotics", "Model Predictive Control", "Soft Robotics" ]
null
4,319
null
null
[ -0.05591654032468796, -0.006417667493224144, -0.04670991376042366, 0.035417258739471436, 0.04974190890789032, 0.006032076198607683, 0.009337084367871284, 0.02208741009235382, -0.030530104413628578, -0.03645945340394974, 0.013064524158835411, -0.006865911651402712, -0.06807127594947815, -0....
Neural Network Approximation based on Hausdorff distance of Tropical Zonotopes
https://openreview.net/forum?id=oiZJwC_fyS
[ "Panagiotis Misiakos", "Georgios Smyrnis", "George Retsinas", "Petros Maragos" ]
Poster
null
In this work we theoretically contribute to neural network approximation by providing a novel tropical geometrical viewpoint to structured neural network compression. In particular, we show that the approximation error between two neural networks with ReLU activations and one hidden layer depends on the Hausdorff dista...
[ "Tropical Geometry", "Zonotopes", "Hausdorff Approximation", "Neural Network Compression" ]
null
4,300
null
null
[ -0.03278247267007828, 0.0009324611164629459, -0.02535967156291008, 0.03975444287061691, 0.03792065009474754, 0.07321549952030182, 0.018251726403832436, -0.019985005259513855, -0.04691019281744957, -0.03075268305838108, -0.014146828092634678, -0.010470732115209103, -0.042330749332904816, -0...
Learning Towards The Largest Margins
https://openreview.net/forum?id=hqkhcFHOeKD
[ "Xiong Zhou", "Xianming Liu", "Deming Zhai", "Junjun Jiang", "Xin Gao", "Xiangyang Ji" ]
Poster
null
One of the main challenges for feature representation in deep learning-based classification is the design of appropriate loss functions that exhibit strong discriminative power. The classical softmax loss does not explicitly encourage discriminative learning of features. A popular direction of research is to incorporat...
[ "loss function design", "margin-based loss", "classification" ]
null
4,295
2206.11589
title_snapshot
[ -0.01920790784060955, -0.011458579450845718, 0.018242759630084038, 0.010856067761778831, 0.022024134173989296, -0.0033956083934754133, 0.0028417406138032675, -0.022787000983953476, -0.018495382741093636, -0.03962497040629387, -0.003740292973816395, -0.005502191837877035, -0.07525915652513504...
Adversarial Support Alignment
https://openreview.net/forum?id=26gKg6x-ie
[ "Shangyuan Tong", "Timur Garipov", "Yang Zhang", "Shiyu Chang", "Tommi S. Jaakkola" ]
Spotlight
null
We study the problem of aligning the supports of distributions. Compared to the existing work on distribution alignment, support alignment does not require the densities to be matched. We propose symmetric support difference as a divergence measure to quantify the mismatch between supports. We show that select discrimi...
[ "support alignment", "distribution alignment", "optimal transport", "domain adaptation" ]
null
4,292
2203.08908
title_snapshot
[ 0.012262252159416676, -0.033476389944553375, -0.04003881290555, 0.03661813586950302, 0.024399789050221443, 0.039811208844184875, 0.009693868458271027, -0.021056005731225014, -0.012394878081977367, -0.04635113477706909, 0.010078883729875088, 0.025163259357213974, -0.09092910587787628, -0.01...
Patch-Fool: Are Vision Transformers Always Robust Against Adversarial Perturbations?
https://openreview.net/forum?id=28ib9tf6zhr
[ "Yonggan Fu", "Shunyao Zhang", "Shang Wu", "Cheng Wan", "Yingyan Lin" ]
Poster
null
Vision transformers (ViTs) have recently set off a new wave in neural architecture design thanks to their record-breaking performance in various vision tasks. In parallel, to fulfill the goal of deploying ViTs into real-world vision applications, their robustness against potential malicious attacks has gained increasin...
[ "Vision transformer", "adversarial examples", "robustness" ]
null
4,289
2203.08392
title_snapshot
[ 0.018151002004742622, -0.035827428102493286, 0.012355857528746128, 0.0405123233795166, 0.020372843369841576, 0.03007025644183159, 0.034725356847047806, 0.002804548479616642, -0.0018431966891512275, -0.050656769424676895, -0.019618479534983635, 0.028321098536252975, -0.07116739451885223, -0...
AdaMatch: A Unified Approach to Semi-Supervised Learning and Domain Adaptation
https://openreview.net/forum?id=Q5uh1Nvv5dm
[ "David Berthelot", "Rebecca Roelofs", "Kihyuk Sohn", "Nicholas Carlini", "Alexey Kurakin" ]
Poster
null
We extend semi-supervised learning to the problem of domain adaptation to learn significantly higher-accuracy models that train on one data distribution and test on a different one. With the goal of generality, we introduce AdaMatch, a unified solution for unsupervised domain adaptation (UDA), semi-supervised learning ...
[ "unsupervised domain adaptation", "semi-supervised learning", "semi-supervised domain adaptation" ]
null
4,287
2106.04732
title_snapshot
[ -0.017769213765859604, -0.05571966618299484, -0.02797415293753147, 0.051120445132255554, 0.03844357654452324, 0.012102008797228336, 0.04170062020421028, -0.003482941072434187, 0.009740985929965973, -0.025732893496751785, -0.04368819296360016, 0.008893475867807865, -0.07952787727117538, 0.0...
Complete Verification via Multi-Neuron Relaxation Guided Branch-and-Bound
https://openreview.net/forum?id=l_amHf1oaK
[ "Claudio Ferrari", "Mark Niklas Mueller", "Nikola Jovanović", "Martin Vechev" ]
Poster
null
State-of-the-art neural network verifiers are fundamentally based on one of two paradigms: either encoding the whole verification problem via tight multi-neuron convex relaxations or applying a Branch-and-Bound (BaB) procedure leveraging imprecise but fast bounding methods on a large number of easier subproblems. The f...
[ "Certified Robustness", "Branch-and-Bound", "Convex Relaxation" ]
null
4,286
2205.00263
title_snapshot
[ -0.021137785166502, 0.0047453707084059715, -0.0010354052064940333, 0.03170888125896454, 0.032064687460660934, 0.040621478110551834, 0.001633001142181456, -0.0016707131871953607, -0.03977632895112038, -0.009139111265540123, -0.0102770971134305, 0.00302078016102314, -0.04153803363442421, -0....
GreaseLM: Graph REASoning Enhanced Language Models
https://openreview.net/forum?id=41e9o6cQPj
[ "Xikun Zhang", "Antoine Bosselut", "Michihiro Yasunaga", "Hongyu Ren", "Percy Liang", "Christopher D Manning", "Jure Leskovec" ]
Spotlight
null
Answering complex questions about textual narratives requires reasoning over both stated context and the world knowledge that underlies it. However, pretrained language models (LM), the foundation of most modern QA systems, do not robustly represent latent relationships between concepts, which is necessary for reasonin...
[ "language models", "commonsense", "question answering", "knowledge graphs", "KG augmentation" ]
null
4,280
null
null
[ 0.00640596030279994, 0.017063196748495102, 0.013645059429109097, 0.045069918036460876, 0.05242433398962021, -0.00297692627646029, 0.03075293079018593, 0.01381136104464531, -0.01240513939410448, 0.016665123403072357, 0.005095789674669504, 0.077324278652668, -0.06678563356399536, -0.01407518...
Learning Fast Samplers for Diffusion Models by Differentiating Through Sample Quality
https://openreview.net/forum?id=VFBjuF8HEp
[ "Daniel Watson", "William Chan", "Jonathan Ho", "Mohammad Norouzi" ]
Poster
null
Diffusion models have emerged as an expressive family of generative models rivaling GANs in sample quality and autoregressive models in likelihood scores. Standard diffusion models typically require hundreds of forward passes through the model to generate a single high-fidelity sample. We introduce Differentiable Diffu...
[]
null
4,276
2202.05830
title_snapshot
[ -0.011732886545360088, -0.02477891929447651, 0.0008653641561977565, 0.07288917154073715, 0.05641192942857742, 0.04173620045185089, 0.018127787858247757, 0.00046509236562997103, -0.005634235683828592, -0.07702889293432236, 0.022723866626620293, -0.020497746765613556, -0.04182175174355507, 0...
Distribution Compression in Near-Linear Time
https://openreview.net/forum?id=lzupY5zjaU9
[ "Abhishek Shetty", "Raaz Dwivedi", "Lester Mackey" ]
Poster
null
In distribution compression, one aims to accurately summarize a probability distribution $\mathbb{P}$ using a small number of representative points. Near-optimal thinning procedures achieve this goal by sampling $n$ points from a Markov chain and identifying $\sqrt{n}$ points with $\widetilde{\mathcal{O}}(1/\sqrt{n})$ ...
[ "Distribution compression", "linear time", "thinning", "i.i.d. sampling", "Markov chain Monte Carlo", "maximum mean discrepancy", "reproducing kernel Hilbert space" ]
null
4,274
2111.07941
title_snapshot
[ -0.0065964506939053535, -0.017148630693554878, 0.01704338565468788, 0.02955245040357113, 0.05752433463931084, 0.07465869933366776, -0.00476867938414216, -0.020643403753638268, -0.02699916809797287, -0.060257717967033386, -0.010760365054011345, -0.023930398747324944, -0.04828495532274246, 0...
Capturing Structural Locality in Non-parametric Language Models
https://openreview.net/forum?id=nnU3IUMJmN
[ "Frank F. Xu", "Junxian He", "Graham Neubig", "Vincent Josua Hellendoorn" ]
Poster
null
Structural locality is a ubiquitous feature of real-world datasets, wherein data points are organized into local hierarchies. Some examples include topical clusters in text or project hierarchies in source code repositories. In this paper, we explore utilizing this structural locality within non-parametric language mod...
[]
null
4,271
2110.02870
title_snapshot
[ -0.023929933086037636, -0.005728657823055983, -0.02763332612812519, 0.06782840937376022, 0.023808101192116737, 0.05657760798931122, 0.038090456277132034, 0.0060545941814780235, -0.015722444280982018, -0.024241145700216293, 0.0002105647581629455, -0.01365828700363636, -0.08186227828264236, ...
Audio Lottery: Speech Recognition Made Ultra-Lightweight, Noise-Robust, and Transferable
https://openreview.net/forum?id=9Nk6AJkVYB
[ "Shaojin Ding", "Tianlong Chen", "Zhangyang Wang" ]
Poster
null
Lightweight speech recognition models have seen explosive demands owing to a growing amount of speech-interactive features on mobile devices. Since designing such systems from scratch is non-trivial, practitioners typically choose to compress large (pre-trained) speech models. Recently, lottery ticket hypothesis reveal...
[ "Speech Recognition", "Lottery Ticket Hypothesis" ]
null
4,269
null
null
[ -0.0402691513299942, -0.029688699170947075, -0.022530077025294304, 0.02956710010766983, 0.022242698818445206, 0.052523836493492126, 0.015743842348456383, 0.028249232098460197, -0.022634757682681084, -0.057231511920690536, 0.01194078754633665, 0.016808589920401573, -0.05690649524331093, 0.0...
Learning meta-features for AutoML
https://openreview.net/forum?id=DTkEfj0Ygb8
[ "Herilalaina Rakotoarison", "Louisot Milijaona", "Andry RASOANAIVO", "Michele Sebag", "Marc Schoenauer" ]
Spotlight
null
This paper tackles the AutoML problem, aimed to automatically select an ML algorithm and its hyper-parameter configuration most appropriate to the dataset at hand. The proposed approach, MetaBu, learns new meta-features via an Optimal Transport procedure, aligning the manually designed \mf s with the space of distribut...
[ "AutoML", "Meta-features", "Hyper-parameter Optimization", "Optimal Transport" ]
null
4,263
null
null
[ -0.009407934732735157, -0.02653425559401512, 0.011350702494382858, 0.027819225564599037, 0.02054225467145443, 0.026580877602100372, 0.034192539751529694, -0.02793271839618683, 0.0024645659141242504, -0.019553618505597115, -0.037598349153995514, 0.007552155293524265, -0.08995380997657776, 0...
Minibatch vs Local SGD with Shuffling: Tight Convergence Bounds and Beyond
https://openreview.net/forum?id=LdlwbBP2mlq
[ "Chulhee Yun", "Shashank Rajput", "Suvrit Sra" ]
Oral
null
In distributed learning, local SGD (also known as federated averaging) and its simple baseline minibatch SGD are widely studied optimization methods. Most existing analyses of these methods assume independent and unbiased gradient estimates obtained via with-replacement sampling. In contrast, we study shuffling-based v...
[ "Local SGD", "Minibatch SGD", "Shuffling", "Without-replacement", "Convex Optimization", "Stochastic Optimization", "Federated Learning", "Large Scale Learning", "Distributed Learning" ]
null
4,253
2110.10342
title_snapshot
[ -0.02899247594177723, -0.05855562165379524, 0.0019637057557702065, 0.058686595410108566, 0.02564191445708275, 0.033850595355033875, 0.0354900024831295, -0.015847204253077507, -0.009652083739638329, -0.06165337935090065, 0.021737586706876755, -0.04224560409784317, -0.06738876551389694, -0.0...
Learning to Map for Active Semantic Goal Navigation
https://openreview.net/forum?id=swrMQttr6wN
[ "Georgios Georgakis", "Bernadette Bucher", "Karl Schmeckpeper", "Siddharth Singh", "Kostas Daniilidis" ]
Poster
null
We consider the problem of object goal navigation in unseen environments. Solving this problem requires learning of contextual semantic priors, a challenging endeavour given the spatial and semantic variability of indoor environments. Current methods learn to implicitly encode these priors through goal-oriented navigat...
[ "visual navigation", "semantic map", "uncertainty estimation" ]
null
4,251
2106.15648
title_snapshot
[ -0.01113781239837408, 0.009038056246936321, 0.013009276241064072, 0.009681624360382557, 0.03258030489087105, 0.019777940586209297, 0.02015347220003605, 0.008595659397542477, -0.046992987394332886, -0.03862784430384636, -0.07521676272153854, 0.0005814702599309385, -0.049082498997449875, -0....
Benchmarking the Spectrum of Agent Capabilities
https://openreview.net/forum?id=1W0z96MFEoH
[ "Danijar Hafner" ]
Poster
null
Evaluating the general abilities of intelligent agents requires complex simulation environments. Existing benchmarks typically evaluate only one narrow task per environment, requiring researchers to perform expensive training runs on many different environments. We introduce Crafter, an open world survival game with vi...
[ "Evaluation", "Reinforcement Learning", "Environment", "Benchmark", "Unsupervised Reinforcement Learning", "Exploration" ]
null
4,249
2109.06780
title_snapshot
[ -0.02538454532623291, -0.01864846795797348, 0.010548566468060017, 0.04292207583785057, 0.050442613661289215, 0.009944385848939419, 0.033179569989442825, 0.011140881106257439, -0.031592369079589844, -0.046720705926418304, -0.028393292799592018, 0.029952462762594223, -0.054250359535217285, -...
Mind the Gap: Domain Gap Control for Single Shot Domain Adaptation for Generative Adversarial Networks
https://openreview.net/forum?id=vqGi8Kp0wM
[ "Peihao Zhu", "Rameen Abdal", "John Femiani", "Peter Wonka" ]
Poster
null
We present a new method for one shot domain adaptation. The input to our method is trained GAN that can produce images in domain A and a single reference image I_B from domain B. The proposed algorithm can translate any output of the trained GAN from domain A to domain B. There are two main advantages of our method com...
[ "GAN", "StyleGAN", "Clip", "Domain Adaptation", "Style Transfer", "Single Shot" ]
null
4,248
2110.08398
title_snapshot
[ -0.006750038359314203, -0.0025387706700712442, -0.0019621537066996098, 0.042974431067705154, 0.036857184022665024, -0.000959891127422452, 0.033923108130693436, -0.0062073892913758755, -0.006206844933331013, -0.043452534824609756, -0.025677310302853584, 0.002621937543153763, -0.07502660155296...
The Hidden Convex Optimization Landscape of Regularized Two-Layer ReLU Networks: an Exact Characterization of Optimal Solutions
https://openreview.net/forum?id=Z7Lk2cQEG8a
[ "Yifei Wang", "Jonathan Lacotte", "Mert Pilanci" ]
Oral
null
We prove that finding all globally optimal two-layer ReLU neural networks can be performed by solving a convex optimization program with cone constraints. Our analysis is novel, characterizes all optimal solutions, and does not leverage duality-based analysis which was recently used to lift neural network training into...
[ "Neural networks", "global optimization", "convex optimization", "convex analysis" ]
null
4,247
2006.05900
title_judge
[ -0.0379008911550045, -0.011416534893214703, -0.002094645518809557, 0.0352158322930336, 0.033871330320835114, 0.05276697501540184, 0.02014085091650486, 0.016112610697746277, -0.03254618123173714, -0.03458240628242493, -0.017060915008187294, 0.005934346932917833, -0.04126765951514244, -0.033...
On Evaluation Metrics for Graph Generative Models
https://openreview.net/forum?id=EnwCZixjSh
[ "Rylee Thompson", "Boris Knyazev", "Elahe Ghalebi", "Jungtaek Kim", "Graham W. Taylor" ]
Poster
null
In image generation, generative models can be evaluated naturally by visually inspecting model outputs. However, this is not always the case for graph generative models (GGMs), making their evaluation challenging. Currently, the standard process for evaluating GGMs suffers from three critical limitations: i) it does no...
[]
null
4,244
2201.09871
title_snapshot
[ 0.008215300738811493, -0.024695776402950287, 0.01392712164670229, 0.05597800016403198, 0.01914466917514801, 0.02422966994345188, 0.028554996475577354, 0.013650204055011272, -0.0008990559726953506, -0.05586124211549759, -0.012151016853749752, -0.011694323271512985, -0.076881043612957, 0.001...
Selective Ensembles for Consistent Predictions
https://openreview.net/forum?id=HfUyCRBeQc
[ "Emily Black", "Klas Leino", "Matt Fredrikson" ]
Poster
null
Recent work has shown that models trained to the same objective, and which achieve similar measures of accuracy on consistent test data, may nonetheless behave very differently on individual predictions. This inconsistency is undesirable in high-stakes contexts, such as medical diagnosis and finance. We show that this ...
[ "consistency", "prediction consistency", "model duplicity", "inconsistent predictions", "deep models", "deep networks", "explanations", "saliency maps", "gradient-based explanations", "fairness", "interpretability" ]
null
4,240
2111.08230
title_snapshot
[ -0.009761996567249298, -0.005322206299751997, -0.03931880742311478, 0.03711261600255966, 0.053788188844919205, 0.043509066104888916, 0.03606605529785156, -0.008953317068517208, -0.025161484256386757, -0.05286022275686264, -0.008566807024180889, 0.04551095515489578, -0.0880015417933464, -0....
Graph Condensation for Graph Neural Networks
https://openreview.net/forum?id=WLEx3Jo4QaB
[ "Wei Jin", "Lingxiao Zhao", "Shichang Zhang", "Yozen Liu", "Jiliang Tang", "Neil Shah" ]
Poster
null
Given the prevalence of large-scale graphs in real-world applications, the storage and time for training neural models have raised increasing concerns. To alleviate the concerns, we propose and study the problem of graph condensation for graph neural networks (GNNs). Specifically, we aim to condense the large, origina...
[ "data-efficient learning", "graph generation", "graph neural networks" ]
null
4,239
2110.07580
title_snapshot
[ -0.015197329223155975, -0.04533317685127258, 0.014389858581125736, 0.046859268099069595, 0.03605195879936218, 0.04026275873184204, -0.001568431151099503, 0.03395145758986473, -0.017797650769352913, -0.03544633090496063, 0.006938796490430832, -0.026858050376176834, -0.05934195965528488, 0.0...
DIVA: Dataset Derivative of a Learning Task
https://openreview.net/forum?id=bVvMOtLMiw
[ "Yonatan Dukler", "Alessandro Achille", "Giovanni Paolini", "Avinash Ravichandran", "Marzia Polito", "Stefano Soatto" ]
Poster
null
We present a method to compute the derivative of a learning task with respect to a dataset. A learning task is a function from a training set to the validation error, which can be represented by a trained deep neural network (DNN). The ``dataset derivative'' is a linear operator, computed around the trained model, that...
[ "Leave one out cross validation", "AutoML", "dataset optimization" ]
null
4,238
2111.09785
title_snapshot
[ -0.03264626860618591, -0.0325140506029129, -0.03610541671514511, 0.05208342894911766, 0.04917950928211212, 0.03620300069451332, 0.05913151055574417, -0.03941938281059265, -0.0035939819645136595, -0.02371460199356079, -0.013774109072983265, 0.026476996019482613, -0.03674531728029251, 0.0121...
Towards General Function Approximation in Zero-Sum Markov Games
https://openreview.net/forum?id=sA4qIu3zv6v
[ "Baihe Huang", "Jason D. Lee", "Zhaoran Wang", "Zhuoran Yang" ]
Poster
null
This paper considers two-player zero-sum finite-horizon Markov games with simultaneous moves. The study focuses on the challenging settings where the value function or the model is parameterized by general function classes. Provably efficient algorithms for both decoupled and coordinated settings are developed. In the ...
[]
null
4,231
2107.14702
title_snapshot
[ -0.06575582176446915, -0.004751864355057478, 0.017459100112318993, 0.021343868225812912, 0.0508585087954998, 0.03341288864612579, 0.009224921464920044, 0.014797581359744072, -0.023137083277106285, -0.03713100403547287, 0.0006596369785256684, 0.018059631809592247, -0.07858492434024811, 0.00...
Exposing the Implicit Energy Networks behind Masked Language Models via Metropolis--Hastings
https://openreview.net/forum?id=6PvWo1kEvlT
[ "Kartik Goyal", "Chris Dyer", "Taylor Berg-Kirkpatrick" ]
Poster
null
While recent work has shown that scores from models trained by the ubiquitous masked language modeling (MLM) objective effectively discriminate probable from improbable sequences, it is still an open question if these MLMs specify a principled probability distribution over the space of possible sequences. In this paper...
[ "Masked Language Models", "Energy-based models", "Metropolis Hastings Monte Carlo", "Bidirectional Sequence models" ]
null
4,229
2106.02736
title_snapshot
[ -0.0009283504914492369, 0.007165577728301287, -0.019806288182735443, 0.059814486652612686, 0.04996683448553085, 0.02995082549750805, 0.025746850296854973, 0.005293819587677717, -0.02486777864396572, -0.038535792380571365, 0.012602001428604126, 0.025509582832455635, -0.06570160388946533, -0...
ClimateGAN: Raising Climate Change Awareness by Generating Images of Floods
https://openreview.net/forum?id=EZNOb_uNpJk
[ "Victor Schmidt", "Alexandra Luccioni", "Mélisande Teng", "Tianyu Zhang", "Alexia Reynaud", "Sunand Raghupathi", "Gautier Cosne", "Adrien Juraver", "Vahe Vardanyan", "Alex Hernández-García", "Yoshua Bengio" ]
Poster
null
Climate change is a major threat to humanity and the actions required to prevent its catastrophic consequences include changes in both policy-making and individual behaviour. However, taking action requires understanding its seemingly abstract and distant consequences. Projecting the potential impacts of extreme climat...
[ "GAN", "Climate Change", "Domain Adaptation", "Representation Learning", "Computer Vision", "Application" ]
null
4,226
2110.02871
title_snapshot
[ 0.015014139004051685, -0.06922733783721924, -0.005953287705779076, 0.0243765227496624, 0.031412892043590546, 0.012280981987714767, 0.019831256940960884, 0.02756824530661106, -0.03751176595687866, -0.04871542006731033, -0.02974066697061062, -0.021941715851426125, -0.06758061051368713, 0.011...
A Comparison of Hamming Errors of Representative Variable Selection Methods
https://openreview.net/forum?id=nhN-fqxmNGx
[ "Tracy Ke", "Longlin Wang" ]
Poster
null
Lasso is a celebrated method for variable selection in linear models, but it faces challenges when the covariates are moderately or strongly correlated. This motivates alternative approaches such as using a non-convex penalty, adding a ridge regularization, or conducting a post-Lasso thresholding. In this paper, we com...
[ "Lasso", "Hamming error", "phase diagram", "rare and weak signals", "elastic net", "SCAD", "thresholded Lasso", "forward selection", "forward backward selection" ]
null
4,224
2203.15075
title_snapshot
[ -0.02347411960363388, -0.005727654788643122, -0.046594396233558655, 0.005567509680986404, 0.06391686946153641, 0.054932668805122375, 0.03878602012991905, -0.01271058525890112, -0.04298553988337517, -0.05063539743423462, -0.0011496377410367131, -0.0181641336530447, -0.07423166930675507, -0....
Latent Variable Sequential Set Transformers for Joint Multi-Agent Motion Prediction
https://openreview.net/forum?id=Dup_dDqkZC5
[ "Roger Girgis", "Florian Golemo", "Felipe Codevilla", "Martin Weiss", "Jim Aldon D'Souza", "Samira Ebrahimi Kahou", "Felix Heide", "Christopher Pal" ]
Spotlight
null
Robust multi-agent trajectory prediction is essential for the safe control of robotic systems. A major challenge is to efficiently learn a representation that approximates the true joint distribution of contextual, social, and temporal information to enable planning. We propose Latent Variable Sequential Set Transforme...
[ "trajectory prediction", "motion forecasting", "transformers", "latent variable models" ]
null
4,221
2104.00563
title_snapshot
[ 0.02439052052795887, -0.027021506801247597, -0.0055192564614117146, 0.056915100663900375, 0.024895526468753815, 0.03256914019584656, 0.030261022970080376, 0.027445614337921143, -0.02845168486237526, -0.027914267033338547, -0.02717241644859314, 0.005807599518448114, -0.06232992932200432, -0...
A Program to Build E(N)-Equivariant Steerable CNNs
https://openreview.net/forum?id=WE4qe9xlnQw
[ "Gabriele Cesa", "Leon Lang", "Maurice Weiler" ]
Poster
null
Equivariance is becoming an increasingly popular design choice to build data efficient neural networks by exploiting prior knowledge about the symmetries of the problem at hand. Euclidean steerable CNNs are one of the most common classes of equivariant networks. While the constraints these architectures need to satisfy...
[ "equivariance", "3D", "geometric deep learning", "isometries", "steerable CNN" ]
null
4,218
null
null
[ 0.008779685944318771, -0.008286729454994202, 0.0335015133023262, 0.02120264433324337, 0.008856276981532574, -0.00042055852827616036, 0.01165455486625433, 0.030413737520575523, -0.019338825717568398, -0.06113942340016365, -0.026551110669970512, -0.00508848624303937, -0.048564814031124115, 0...
Minimax Optimization with Smooth Algorithmic Adversaries
https://openreview.net/forum?id=UdxJ2fJx7N0
[ "Tanner Fiez", "Chi Jin", "Praneeth Netrapalli", "Lillian J Ratliff" ]
Poster
null
This paper considers minimax optimization $\min_x \max_y f(x, y)$ in the challenging setting where $f$ can be both nonconvex in $x$ and nonconcave in $y$. Though such optimization problems arise in many machine learning paradigms including training generative adversarial networks (GANs) and adversarially robust models,...
[ "Minimax optimization", "two player zero sum games", "generative adversarial networks", "adversarial training" ]
null
4,217
2106.01488
title_snapshot
[ -0.03472980111837387, -0.03452801704406738, 0.010686086490750313, 0.05022333189845085, 0.015900198370218277, 0.03149797022342682, 0.027232134714722633, -0.0014717653393745422, -0.015347708947956562, -0.043776508420705795, -0.020266573876142502, -0.004502099938690662, -0.0599575974047184, 0...
Provably Filtering Exogenous Distractors using Multistep Inverse Dynamics
https://openreview.net/forum?id=RQLLzMCefQu
[ "Yonathan Efroni", "Dipendra Misra", "Akshay Krishnamurthy", "Alekh Agarwal", "John Langford" ]
Oral
null
Many real-world applications of reinforcement learning (RL) require the agent to deal with high-dimensional observations such as those generated from a megapixel camera. Prior work has addressed such problems with representation learning, through which the agent can provably extract endogenous, latent state information...
[ "Reinforcement Learning Theory", "Invariant Representation", "Rich Observation Reinforcement Learning", "Exogenous Noise", "Inverse Dynamics" ]
null
4,212
2110.08847
title_judge
[ -0.026275843381881714, 0.00028070478583686054, -0.02600875496864319, 0.04455571249127388, 0.048680372536182404, 0.015427607111632824, 0.010606436058878899, -0.002070894930511713, -0.044976793229579926, -0.0436888225376606, -0.011295230127871037, -0.0019327023765072227, -0.06812366843223572, ...
On Distributed Adaptive Optimization with Gradient Compression
https://openreview.net/forum?id=CI-xXX9dg9l
[ "Xiaoyun Li", "Belhal Karimi", "Ping Li" ]
Poster
null
We study COMP-AMS, a distributed optimization framework based on gradient averaging and adaptive AMSGrad algorithm. Gradient compression with error feedback is applied to reduce the communication cost in the gradient transmission process. Our convergence analysis of COMP-AMS shows that such compressed gradient averagin...
[]
null
4,202
2205.05632
title_snapshot
[ -0.02305132895708084, -0.026875963434576988, 0.01928786374628544, 0.02342737652361393, 0.04674481600522995, 0.04906865209341049, 0.03522474691271782, -0.01585792750120163, -0.031328558921813965, -0.07774508744478226, -0.02127598226070404, -0.005677167791873217, -0.05569029226899147, -0.002...
Leveraging unlabeled data to predict out-of-distribution performance
https://openreview.net/forum?id=o_HsiMPYh_x
[ "Saurabh Garg", "Sivaraman Balakrishnan", "Zachary Chase Lipton", "Behnam Neyshabur", "Hanie Sedghi" ]
Poster
null
Real-world machine learning deployments are characterized by mismatches between the source (training) and target (test) distributions that may cause performance drops. In this work, we investigate methods for predicting the target domain accuracy using only labeled source data and unlabeled target data. We propose Aver...
[ "Distribution Shift", "OOD error prediction", "Deep Learning" ]
null
4,190
2201.04234
title_snapshot
[ 0.015016753226518631, -0.03810586407780647, -0.008827107958495617, 0.06131862848997116, 0.03311043977737427, -0.007001237943768501, 0.04139753803610802, 0.0031161708757281303, -0.002717345952987671, -0.02617824450135231, 0.008607221767306328, 0.014677110128104687, -0.08624003827571869, -0....
VC dimension of partially quantized neural networks in the overparametrized regime
https://openreview.net/forum?id=7udZAsEzd60
[ "Yutong Wang", "Clayton Scott" ]
Poster
null
Vapnik-Chervonenkis (VC) theory has so far been unable to explain the small generalization error of overparametrized neural networks. Indeed, existing applications of VC theory to large networks obtain upper bounds on VC dimension that are proportional to the number of weights, and for a large class of networks, these ...
[ "VC dimension", "quantized neural networks", "classification", "minimax theory", "overparametrization" ]
null
4,186
2110.02456
title_snapshot
[ -0.05540093779563904, -0.021180998533964157, 0.0018604776123538613, 0.0426468551158905, 0.03279135003685951, 0.039994366466999054, 0.025200286880135536, -0.0006173037691041827, -0.030761756002902985, -0.03916789963841438, 0.0008780244970694184, -0.02731035277247429, -0.05828813463449478, -...
Optimal Representations for Covariate Shift
https://openreview.net/forum?id=Rf58LPCwJj0
[ "Yangjun Ruan", "Yann Dubois", "Chris J. Maddison" ]
Poster
null
Machine learning systems often experience a distribution shift between training and testing. In this paper, we introduce a simple variational objective whose optima are exactly the set of all representations on which risk minimizers are guaranteed to be robust to any distribution shift that preserves the Bayes predicto...
[ "distribution shift", "domain generalization", "representation learning", "self-supervised learning", "invariance", "robustness" ]
null
4,179
2201.00057
title_snapshot
[ -0.0301264226436615, -0.003631342900916934, -0.0046005756594240665, 0.036583900451660156, 0.041838664561510086, 0.05858062952756882, 0.03420409560203552, -0.021110184490680695, -0.0032593458890914917, -0.04388805851340294, -0.028492169454693794, 0.024465089663863182, -0.08057945221662521, ...
Fortuitous Forgetting in Connectionist Networks
https://openreview.net/forum?id=ei3SY1_zYsE
[ "Hattie Zhou", "Ankit Vani", "Hugo Larochelle", "Aaron Courville" ]
Poster
null
Forgetting is often seen as an unwanted characteristic in both human and machine learning. However, we propose that forgetting can in fact be favorable to learning. We introduce forget-and-relearn as a powerful paradigm for shaping the learning trajectories of artificial neural networks. In this process, the forgetting...
[ "Neural Networks", "Generalization", "Iterative Training", "Compositionality", "Iterated Learning" ]
null
4,178
2202.00155
title_snapshot
[ -0.03601858764886856, -0.01638476923108101, -0.0002860444947145879, 0.024938957765698433, 0.04082844406366348, 0.016027048230171204, 0.017533140257000923, 0.0389418751001358, -0.06214981526136398, -0.011816618964076042, -0.015251372009515762, 0.027799008414149284, -0.04913446307182312, -0....
Understanding Latent Correlation-Based Multiview Learning and Self-Supervision: An Identifiability Perspective
https://openreview.net/forum?id=5FUq05QRc5b
[ "Qi Lyu", "Xiao Fu", "Weiran Wang", "Songtao Lu" ]
Spotlight
null
Multiple views of data, both naturally acquired (e.g., image and audio) and artificially produced (e.g., via adding different noise to data samples), have proven useful in enhancing representation learning. Natural views are often handled by multiview analysis tools, e.g., (deep) canonical correlation analysis [(D)CCA]...
[]
null
4,176
2106.07115
title_snapshot
[ 0.03567405790090561, -0.007812573574483395, -0.003468394512310624, 0.04712773114442825, 0.0381760448217392, 0.027708720415830612, 0.028910920023918152, -0.014374040998518467, -0.020068634301424026, -0.049834173172712326, -0.022781020030379295, 0.011280801147222519, -0.06619059294462204, 0....
EigenGame Unloaded: When playing games is better than optimizing
https://openreview.net/forum?id=So6YAqnqgMj
[ "Ian Gemp", "Brian McWilliams", "Claire Vernade", "Thore Graepel" ]
Poster
null
We build on the recently proposed EigenGame that views eigendecomposition as a competitive game. EigenGame's updates are biased if computed using minibatches of data, which hinders convergence and more sophisticated parallelism in the stochastic setting. In this work, we propose an unbiased stochastic update that is as...
[ "pca", "principal components analysis", "nash", "games", "eigendecomposition", "svd", "singular value decomposition" ]
null
4,175
2102.04152
title_snapshot
[ -0.025404557585716248, -0.033955447375774384, 0.029007496312260628, 0.034651827067136765, 0.028808634728193283, 0.03054588846862316, 0.015675213187932968, 0.01863430067896843, -0.025346364825963974, -0.07230395823717117, 0.008160172961652279, -0.007921654731035233, -0.09435790777206421, -0...
Contextualized Scene Imagination for Generative Commonsense Reasoning
https://openreview.net/forum?id=Oh1r2wApbPv
[ "PeiFeng Wang", "Jonathan Zamora", "Junfeng Liu", "Filip Ilievski", "Muhao Chen", "Xiang Ren" ]
Poster
null
Humans use natural language to compose common concepts from their environment into plausible, day-to-day scene descriptions. However, such generative commonsense reasoning (GCSR) skills are lacking in state-of-the-art text generation methods. Descriptive sentences about arbitrary concepts generated by neural text gener...
[ "Commonsense reasoning", "constrained text generation", "knowledge representation" ]
null
4,167
2112.06318
title_snapshot
[ -0.009149855002760887, 0.003954154439270496, 0.0009157940512523055, 0.07476968318223953, 0.04066033661365509, 0.007934866473078728, 0.01334375236183405, 0.02860991843044758, -0.026751438155770302, 0.013016085140407085, -0.05429317429661751, 0.029181892052292824, -0.05537872016429901, 0.002...
Scene Transformer: A unified architecture for predicting future trajectories of multiple agents
https://openreview.net/forum?id=Wm3EA5OlHsG
[ "Jiquan Ngiam", "Vijay Vasudevan", "Benjamin Caine", "Zhengdong Zhang", "Hao-Tien Lewis Chiang", "Jeffrey Ling", "Rebecca Roelofs", "Alex Bewley", "Chenxi Liu", "Ashish Venugopal", "David J Weiss", "Benjamin Sapp", "Zhifeng Chen", "Jonathon Shlens" ]
Poster
null
Predicting the motion of multiple agents is necessary for planning in dynamic environments. This task is challenging for autonomous driving since agents (e.g., vehicles and pedestrians) and their associated behaviors may be diverse and influence one another. Most prior work have focused on predicting independent future...
[ "trajectory prediction", "motion forecasting", "multi-task learning", "attention", "autonomous vehicles" ]
null
4,165
2106.08417
title_judge
[ -0.013004750944674015, -0.022105783224105835, 0.009595716372132301, 0.034346770495176315, 0.031473878771066666, 0.01846453733742237, 0.03170495852828026, 0.021227749064564705, -0.027394287288188934, -0.03162037581205368, -0.02828560583293438, 0.020230302587151527, -0.08159495145082474, -0....
DISSECT: Disentangled Simultaneous Explanations via Concept Traversals
https://openreview.net/forum?id=qY79G8jGsep
[ "Asma Ghandeharioun", "Been Kim", "Chun-Liang Li", "Brendan Jou", "Brian Eoff", "Rosalind Picard" ]
Poster
null
Explaining deep learning model inferences is a promising venue for scientific understanding, improving safety, uncovering hidden biases, evaluating fairness, and beyond, as argued by many scholars. One of the principal benefits of counterfactual explanations is allowing users to explore "what-if" scenarios through what...
[ "Explainability", "Interpretability", "Counterfactual generation", "Generative Adversarial Network", "Variational Autoencoder" ]
null
4,164
2105.15164
title_snapshot
[ -0.001967641059309244, -0.038416821509599686, -0.05772198736667633, 0.0836222916841507, 0.03133886680006981, -0.015353012830018997, 0.011606015264987946, -0.007159179542213678, 0.001259441371075809, -0.0343901626765728, -0.03818608075380325, 0.017767284065485, -0.05713758245110512, 0.03443...