title
stringlengths
15
135
paper_url
stringlengths
45
65
authors
listlengths
1
15
type
stringclasses
0 values
primary_area
stringclasses
0 values
abstract
large_stringlengths
289
1.91k
keywords
listlengths
0
0
TL;DR
large_stringclasses
0 values
submission_number
int64
1
1.08k
arxiv_id
stringlengths
10
10
arxiv_id_source
stringclasses
2 values
embedding
listlengths
768
768
Selective Dyna-Style Planning Under Limited Model Capacity
https://proceedings.mlr.press/v119/abbas20a.html
[ "Zaheer Abbas", "Samuel Sokota", "Erin Talvitie", "Martha White" ]
null
null
In model-based reinforcement learning, planning with an imperfect model of the environment has the potential to harm learning progress. But even when a model is imperfect, it may still contain information that is useful for planning. In this paper, we investigate the idea of using an imperfect model selectively. The ag...
[]
null
1
2007.02418
title_snapshot
[ -0.03979296609759331, -0.011885025538504124, -0.057498347014188766, 0.02720605954527855, 0.05397235229611397, 0.0433514378964901, 0.007139356806874275, -0.029098082333803177, -0.055756211280822754, -0.034103281795978546, -0.02149016037583351, 0.023484310135245323, -0.07693992555141449, -0....
A distributional view on multi-objective policy optimization
https://proceedings.mlr.press/v119/abdolmaleki20a.html
[ "Abbas Abdolmaleki", "Sandy Huang", "Leonard Hasenclever", "Michael Neunert", "Francis Song", "Martina Zambelli", "Murilo Martins", "Nicolas Heess", "Raia Hadsell", "Martin Riedmiller" ]
null
null
Many real-world problems require trading off multiple competing objectives. However, these objectives are often in different units and/or scales, which can make it challenging for practitioners to express numerical preferences over objectives in their native units. In this paper we propose a novel algorithm for multi-o...
[]
null
2
2005.07513
title_snapshot
[ -0.031104527413845062, 0.0006475946865975857, 0.007637488655745983, 0.03774559870362282, 0.02417341247200966, 0.058745961636304855, -0.00919093657284975, -0.02399859018623829, -0.022294320166110992, -0.039940156042575836, 0.0022541205398738384, 0.019623856991529465, -0.08652617037296295, -...
Efficient Optimistic Exploration in Linear-Quadratic Regulators via Lagrangian Relaxation
https://proceedings.mlr.press/v119/abeille20a.html
[ "Marc Abeille", "Alessandro Lazaric" ]
null
null
We study the exploration-exploitation dilemma in the linear quadratic regulator (LQR) setting. Inspired by the extended value iteration algorithm used in optimistic algorithms for finite MDPs, we propose to relax the optimistic optimization of \ofulq and cast it into a constrained \emph{extended} LQR problem, where an ...
[]
null
3
2007.06482
title_snapshot
[ -0.03114425204694271, -0.019292835146188736, 0.005178960971534252, 0.04868016019463539, 0.051395367830991745, 0.02796582691371441, 0.0021623685024678707, 0.0008575823740102351, -0.014623112976551056, -0.04174206778407097, -0.007997890934348106, 0.003188374685123563, -0.07889243960380554, -...
Super-efficiency of automatic differentiation for functions defined as a minimum
https://proceedings.mlr.press/v119/ablin20a.html
[ "Pierre Ablin", "Gabriel Peyré", "Thomas Moreau" ]
null
null
In min-min optimization or max-min optimization, one has to compute the gradient of a function defined as a minimum. In most cases, the minimum has no closed-form, and an approximation is obtained via an iterative algorithm. There are two usual ways of estimating the gradient of the function: using either an analytic f...
[]
null
4
2002.03722
title_snapshot
[ -0.05891513079404831, 0.007307784166187048, 0.016969148069620132, 0.0222121924161911, 0.0422762893140316, 0.03469221666455269, 0.024669140577316284, -0.026782218366861343, -0.026605360209941864, -0.01665789633989334, -0.017578760161995888, 0.008930043317377567, -0.0371948778629303, 0.01210...
A Geometric Approach to Archetypal Analysis via Sparse Projections
https://proceedings.mlr.press/v119/abrol20a.html
[ "Vinayak Abrol", "Pulkit Sharma" ]
null
null
Archetypal analysis (AA) aims to extract patterns using self-expressive decomposition of data as convex combinations of extremal points (on the convex hull) of the data. This work presents a computationally efficient greedy AA (GAA) algorithm. GAA leverages the underlying geometry of AA, is scalable to larger datasets,...
[]
null
5
null
null
[ 0.005312369205057621, 0.003400438232347369, -0.023876314982771873, 0.02008054032921791, 0.019686810672283173, 0.03411088511347771, 0.014352180063724518, -0.04901104420423508, -0.03779435157775879, -0.05894380807876587, -0.018636085093021393, 0.005921488162130117, -0.08456029742956161, 0.00...
Context Aware Local Differential Privacy
https://proceedings.mlr.press/v119/acharya20a.html
[ "Jayadev Acharya", "Kallista Bonawitz", "Peter Kairouz", "Daniel Ramage", "Ziteng Sun" ]
null
null
Local differential privacy (LDP) is a strong notion of privacy that often leads to a significant drop in utility. The original definition of LDP assumes that all the elements in the data domain are equally sensitive. However, in many real-life applications, some elements are more sensitive than others. We propose a con...
[]
null
6
1911.00038
title_snapshot
[ -0.0108714010566473, 0.020418066531419754, -0.00029546479345299304, 0.07250507920980453, 0.05235564336180687, 0.022454869002103806, 0.027497539296746254, -0.03577994182705879, 0.00900258682668209, -0.023159250617027283, -0.0006873104139231145, -0.03472021222114563, -0.05891593545675278, -0...
Efficient Intervention Design for Causal Discovery with Latents
https://proceedings.mlr.press/v119/addanki20a.html
[ "Raghavendra Addanki", "Shiva Kasiviswanathan", "Andrew Mcgregor", "Cameron Musco" ]
null
null
We consider recovering a causal graph in presence of latent variables, where we seek to minimize the cost of interventions used in the recovery process. We consider two intervention cost models: (1) a linear cost model where the cost of an intervention on a subset of variables has a linear form, and (2) an identity cos...
[]
null
7
2005.11736
title_snapshot
[ -0.015009771101176739, -0.01844882406294346, -0.02572914958000183, 0.031185612082481384, 0.04000372812151909, 0.030493415892124176, 0.025579597800970078, 0.0071796695701777935, 0.005819897633045912, -0.057807017117738724, 0.005565491504967213, -0.011420977301895618, -0.037611283361911774, ...
The Neural Tangent Kernel in High Dimensions: Triple Descent and a Multi-Scale Theory of Generalization
https://proceedings.mlr.press/v119/adlam20a.html
[ "Ben Adlam", "Jeffrey Pennington" ]
null
null
Modern deep learning models employ considerably more parameters than required to fit the training data. Whereas conventional statistical wisdom suggests such models should drastically overfit, in practice these models generalize remarkably well. An emerging paradigm for describing this unexpected behavior is in terms o...
[]
null
8
2008.06786
title_snapshot
[ -0.05150793120265007, -0.031040487810969353, 0.025256739929318428, 0.033185284584760666, 0.01838214136660099, 0.03459550067782402, 0.017156114801764488, 0.005725111346691847, -0.03224062919616699, -0.040492553263902664, -0.009307395666837692, 0.01346735842525959, -0.039892759174108505, 0.0...
Rank Aggregation from Pairwise Comparisons in the Presence of Adversarial Corruptions
https://proceedings.mlr.press/v119/agarwal20a.html
[ "Arpit Agarwal", "Shivani Agarwal", "Sanjeev Khanna", "Prathamesh Patil" ]
null
null
Rank aggregation from pairwise preferences has widespread applications in recommendation systems and information retrieval. Given the enormous economic and societal impact of these applications, and the consequent incentives for malicious players to manipulate ranking outcomes in their favor, an important challenge is ...
[]
null
9
null
null
[ -0.01812366582453251, -0.025924112647771835, 0.004572195000946522, 0.051023077219724655, 0.01641261950135231, -0.016214700415730476, 0.017670946195721626, -0.01852443441748619, -0.03920890763401985, -0.024413811042904854, -0.02800820767879486, 0.013710018247365952, -0.08048602938652039, -0...
Boosting for Control of Dynamical Systems
https://proceedings.mlr.press/v119/agarwal20b.html
[ "Naman Agarwal", "Nataly Brukhim", "Elad Hazan", "Zhou Lu" ]
null
null
We study the question of how to aggregate controllers for dynamical systems in order to improve their performance. To this end, we propose a framework of boosting for online control. Our main result is an efficient boosting algorithm that combines weak controllers into a provably more accurate one. Empirical evaluation...
[]
null
10
1906.08720
title_snapshot
[ -0.04696071892976761, -0.031426213681697845, -0.008134790696203709, 0.028150709345936775, 0.02655908092856407, -0.0029331070836633444, 0.015359892509877682, 0.007957088761031628, -0.023268602788448334, -0.03239543363451958, 0.017885198816657066, 0.0024026273749768734, -0.09580380469560623, ...
An Optimistic Perspective on Offline Reinforcement Learning
https://proceedings.mlr.press/v119/agarwal20c.html
[ "Rishabh Agarwal", "Dale Schuurmans", "Mohammad Norouzi" ]
null
null
Off-policy reinforcement learning (RL) using a fixed offline dataset of logged interactions is an important consideration in real world applications. This paper studies offline RL using the DQN replay dataset comprising the entire replay experience of a DQN agent on 60 Atari 2600 games. We demonstrate that recent off-p...
[]
null
11
1907.04543
title_snapshot
[ -0.030997024849057198, -0.038085728883743286, -0.0029352428391575813, 0.057221099734306335, 0.033431924879550934, 0.01644369214773178, 0.009018899872899055, 0.029711689800024033, -0.04951184615492821, -0.03206600993871689, -0.04265643283724785, 0.036742642521858215, -0.09704645723104477, -...
Optimal Bounds between f-Divergences and Integral Probability Metrics
https://proceedings.mlr.press/v119/agrawal20a.html
[ "Rohit Agrawal", "Thibaut Horel" ]
null
null
The families of f-divergences (e.g. the Kullback-Leibler divergence) and Integral Probability Metrics (e.g. total variation distance or maximum mean discrepancies) are commonly used in optimization and estimation. In this work, we systematically study the relationship between these two families from the perspective of ...
[]
null
12
2006.05973
title_snapshot
[ -0.05186330899596214, 0.014344142749905586, 0.011072344146668911, 0.013728058896958828, 0.04885846748948097, 0.02883385866880417, 0.025235988199710846, 0.0029368798714131117, -0.02724774368107319, -0.03113032877445221, -0.030430646613240242, 0.0018246894469484687, -0.04847656562924385, -0....
LazyIter: A Fast Algorithm for Counting Markov Equivalent DAGs and Designing Experiments
https://proceedings.mlr.press/v119/ahmaditeshnizi20a.html
[ "Ali Ahmaditeshnizi", "Saber Salehkaleybar", "Negar Kiyavash" ]
null
null
The causal relationships among a set of random variables are commonly represented by a Directed Acyclic Graph (DAG), where there is a directed edge from variable $X$ to variable $Y$ if $X$ is a direct cause of $Y$. From the purely observational data, the true causal graph can be identified up to a Markov Equivalence Cl...
[]
null
13
2006.09670
title_snapshot
[ -0.018901050090789795, -0.023412249982357025, -0.03802867233753204, 0.02210192382335663, 0.04583675041794777, 0.027638835832476616, 0.05004100874066353, 0.00044450987479649484, -0.006786128506064415, -0.03244791179895401, 0.015468346886336803, 0.012480376288294792, -0.07264525443315506, -0...
Learning What to Defer for Maximum Independent Sets
https://proceedings.mlr.press/v119/ahn20a.html
[ "Sungsoo Ahn", "Younggyo Seo", "Jinwoo Shin" ]
null
null
Designing efficient algorithms for combinatorial optimization appears ubiquitously in various scientific fields. Recently, deep reinforcement learning (DRL) frameworks have gained considerable attention as a new approach: they can automate the design of a solver while relying less on sophisticated domain knowledge of t...
[]
null
14
2006.09607
title_snapshot
[ -0.02493409253656864, -0.03728226572275162, -0.012640489265322685, 0.04753340408205986, 0.041024480015039444, 0.024730628356337547, 0.018314404413104057, -0.004670888651162386, -0.010117842815816402, -0.032336317002773285, 0.0013084553647786379, 0.015552469529211521, -0.047559503465890884, ...
Invariant Risk Minimization Games
https://proceedings.mlr.press/v119/ahuja20a.html
[ "Kartik Ahuja", "Karthikeyan Shanmugam", "Kush Varshney", "Amit Dhurandhar" ]
null
null
The standard risk minimization paradigm of machine learning is brittle when operating in environments whose test distributions are different from the training distribution due to spurious correlations. Training on data from many environments and finding invariant predictors reduces the effect of spurious features by co...
[]
null
15
2002.04692
title_snapshot
[ -0.03662508353590965, -0.020193157717585564, 0.011190548539161682, 0.05092943459749222, 0.0285707488656044, 0.03154575079679489, 0.011573957279324532, -0.0038858584593981504, -0.041828371584415436, -0.03506164252758026, -0.03657937049865723, 0.02457396499812603, -0.07831217348575592, -0.00...
Why bigger is not always better: on finite and infinite neural networks
https://proceedings.mlr.press/v119/aitchison20a.html
[ "Laurence Aitchison" ]
null
null
Recent work has argued that neural networks can be understood theoretically by taking the number of channels to infinity, at which point the outputs become Gaussian process (GP) distributed. However, we note that infinite Bayesian neural networks lack a key facet of the behaviour of real neural networks: the fixed kern...
[]
null
16
1910.08013
title_snapshot
[ -0.04030967876315117, -0.026030566543340683, 0.0004969769506715238, 0.05097297206521034, 0.05335718020796776, 0.03701533377170563, 0.012174168601632118, 0.012747485190629959, -0.031522080302238464, -0.05233456939458847, 0.00556428637355566, 0.008184157311916351, -0.06881439685821533, 0.029...
Discriminative Jackknife: Quantifying Uncertainty in Deep Learning via Higher-Order Influence Functions
https://proceedings.mlr.press/v119/alaa20a.html
[ "Ahmed Alaa", "Mihaela Van Der Schaar" ]
null
null
Deep learning models achieve high predictive accuracy across a broad spectrum of tasks, but rigorously quantifying their predictive uncertainty remains challenging. Usable estimates of predictive uncertainty should (1) cover the true prediction targets with high probability, and (2) discriminate between high- and low c...
[]
null
17
2007.13481
title_snapshot
[ 0.021598121151328087, -0.008684658445417881, -0.04306933656334877, 0.025915715843439102, 0.03187920153141022, 0.03508477658033371, 0.02664564736187458, -0.00845374632626772, -0.01823635771870613, -0.06162667274475098, -0.009132089093327522, 0.03521902486681938, -0.07142505794763565, -0.004...
Frequentist Uncertainty in Recurrent Neural Networks via Blockwise Influence Functions
https://proceedings.mlr.press/v119/alaa20b.html
[ "Ahmed Alaa", "Mihaela Van Der Schaar" ]
null
null
Recurrent neural networks (RNNs) are instrumental in modelling sequential and time-series data. Yet, when using RNNs to inform decision-making, predictions by themselves are not sufficient {—} we also need estimates of predictive uncertainty. Existing approaches for uncertainty quantification in RNNs are based predomin...
[]
null
18
2006.13707
title_snapshot
[ 0.012940564192831516, -0.016214735805988312, -0.04017966240644455, 0.027402585372328758, 0.0668272003531456, 0.05501680448651314, 0.046787433326244354, 0.010824481956660748, -0.039568305015563965, -0.05623305216431618, -0.016051549464464188, 0.000919184647500515, -0.06143125519156456, -0.0...
Random extrapolation for primal-dual coordinate descent
https://proceedings.mlr.press/v119/alacaoglu20a.html
[ "Ahmet Alacaoglu", "Olivier Fercoq", "Volkan Cevher" ]
null
null
We introduce a randomly extrapolated primal-dual coordinate descent method that adapts to sparsity of the data matrix and the favorable structures of the objective function. Our method updates only a subset of primal and dual variables with sparse data, and it uses large step sizes with dense data, retaining the benefi...
[]
null
19
2007.06528
title_snapshot
[ -0.024802885949611664, -0.008946801535785198, 0.0227873083204031, 0.06078679487109184, 0.049282900989055634, 0.05651012063026428, -0.012923593632876873, -0.012954434379935265, -0.031593021005392075, -0.04904179647564888, -0.03653072193264961, -0.00487954868003726, -0.056721314787864685, -0...
A new regret analysis for Adam-type algorithms
https://proceedings.mlr.press/v119/alacaoglu20b.html
[ "Ahmet Alacaoglu", "Yura Malitsky", "Panayotis Mertikopoulos", "Volkan Cevher" ]
null
null
In this paper, we focus on a theory-practice gap for Adam and its variants (AMSGrad, AdamNC, etc.). In practice, these algorithms are used with a constant first-order moment parameter $\beta_{1}$ (typically between $0.9$ and $0.99$). In theory, regret guarantees for online convex optimization require a rapidly decaying...
[]
null
20
2003.09729
title_snapshot
[ -0.03896639123558998, -0.017265452072024345, -0.01297158095985651, 0.03817210718989372, 0.016285395249724388, 0.04963774234056473, 0.03599847853183746, 0.023222237825393677, 0.013387400656938553, -0.05549802631139755, -0.016662003472447395, -0.00023739498283248395, -0.06752508133649826, -0...
Restarted Bayesian Online Change-point Detector achieves Optimal Detection Delay
https://proceedings.mlr.press/v119/alami20a.html
[ "Reda Alami", "Odalric Maillard", "Raphael Feraud" ]
null
null
we consider the problem of sequential change-point detection where both the change-points and the distributions before and after the change are assumed to be unknown. For this problem of primary importance in statistical and sequential learning theory, we derive a variant of the Bayesian Online Change Point Detector p...
[]
null
21
null
null
[ -0.02279072254896164, -0.03188269957900047, -0.001649093464948237, 0.01176220178604126, 0.05428218096494675, 0.028077146038413048, 0.028947805985808372, 0.025384847074747086, -0.029168112203478813, -0.043455254286527634, -0.014390277676284313, 0.010013085789978504, -0.03752017393708229, -0...
Maximum Likelihood with Bias-Corrected Calibration is Hard-To-Beat at Label Shift Adaptation
https://proceedings.mlr.press/v119/alexandari20a.html
[ "Amr Alexandari", "Anshul Kundaje", "Avanti Shrikumar" ]
null
null
Label shift refers to the phenomenon where the prior class probability p(y) changes between the training and test distributions, while the conditional probability p(x|y) stays fixed. Label shift arises in settings like medical diagnosis, where a classifier trained to predict disease given symptoms must be adapted to sc...
[]
null
22
1901.06852
title_snapshot
[ -0.01875411719083786, -0.016590038314461708, -0.021248893812298775, 0.0210763867944479, 0.04936316981911659, 0.019771747291088104, 0.02881282940506935, 0.004810859449207783, -0.03703564405441284, -0.048037558794021606, 0.004763731732964516, 0.03431813418865204, -0.052244000136852264, 0.000...
The Implicit Regularization of Stochastic Gradient Flow for Least Squares
https://proceedings.mlr.press/v119/ali20a.html
[ "Alnur Ali", "Edgar Dobriban", "Ryan Tibshirani" ]
null
null
We study the implicit regularization of mini-batch stochastic gradient descent, when applied to the fundamental problem of least squares regression. We leverage a continuous-time stochastic differential equation having the same moments as stochastic gradient descent, which we call stochastic gradient flow. We give a bo...
[]
null
23
2003.07802
title_snapshot
[ -0.03622846305370331, -0.013987972401082516, 0.0036816985812038183, 0.0205316673964262, 0.03162439540028572, 0.07056320458650589, 0.03299111872911453, 0.0028270967304706573, -0.02270972914993763, -0.03012307547032833, -0.029960863292217255, -0.004995197523385286, -0.056334108114242554, 0.0...
Structural Language Models of Code
https://proceedings.mlr.press/v119/alon20a.html
[ "Uri Alon", "Roy Sadaka", "Omer Levy", "Eran Yahav" ]
null
null
We address the problem of any-code completion - generating a missing piece of source code in a given program without any restriction on the vocabulary or structure. We introduce a new approach to any-code completion that leverages the strict syntax of programming languages to model a code snippet as a tree - structural...
[]
null
24
1910.00577
title_snapshot
[ -0.0038274747785180807, -0.02475312352180481, -0.05043734982609749, 0.03394392877817154, 0.03343033790588379, 0.028341857716441154, 0.04436983913183212, 0.023595109581947327, -0.03255021199584007, -0.02024020254611969, -0.028913989663124084, 0.007192753721028566, -0.07868187129497528, -0.0...
LowFER: Low-rank Bilinear Pooling for Link Prediction
https://proceedings.mlr.press/v119/amin20a.html
[ "Saadullah Amin", "Stalin Varanasi", "Katherine Ann Dunfield", "Günter Neumann" ]
null
null
Knowledge graphs are incomplete by nature, with only a limited number of observed facts from the world knowledge being represented as structured relations between entities. To partly address this issue, an important task in statistical relational learning is that of link prediction or knowledge graph completion. Both l...
[]
null
25
2008.10858
title_snapshot
[ -0.018572963774204254, -0.03220604732632637, 0.015686597675085068, 0.03735686466097832, 0.027355941012501717, -0.0026613303925842047, 0.011617178097367287, -0.00198741490021348, -0.00031634303741157055, -0.027835030108690262, -0.009078511036932468, 0.02839636616408825, -0.06579193472862244, ...
Discount Factor as a Regularizer in Reinforcement Learning
https://proceedings.mlr.press/v119/amit20a.html
[ "Ron Amit", "Ron Meir", "Kamil Ciosek" ]
null
null
Specifying a Reinforcement Learning (RL) task involves choosing a suitable planning horizon, which is typically modeled by a discount factor. It is known that applying RL algorithms with a lower discount factor can act as a regularizer, improving performance in the limited data regime. Yet the exact nature of this regu...
[]
null
26
2007.02040
title_snapshot
[ -0.02496921643614769, -0.04383369907736778, -0.0179374348372221, 0.03680501878261566, 0.05335924029350281, 0.024849090725183487, 0.012232715263962746, 0.003486867994070053, -0.03630934655666351, -0.04520163685083389, -0.0030094310641288757, 0.021184181794524193, -0.03564385697245598, -0.00...
Neuro-Symbolic Visual Reasoning: Disentangling "Visual" from "Reasoning"
https://proceedings.mlr.press/v119/amizadeh20a.html
[ "Saeed Amizadeh", "Hamid Palangi", "Alex Polozov", "Yichen Huang", "Kazuhito Koishida" ]
null
null
Visual reasoning tasks such as visual question answering (VQA) require an interplay of visual perception with reasoning about the question semantics grounded in perception. However, recent advances in this area are still primarily driven by perception improvements (e.g. scene graph generation) rather than reasoning. Ne...
[]
null
27
2006.11524
title_snapshot
[ 0.02679801918566227, -0.0012273287866264582, 0.015013523399829865, 0.05253797769546509, 0.03444330766797066, 0.036257825791835785, 0.023883987218141556, 0.010068791918456554, -0.05730262026190758, -0.02048822119832039, -0.03954111412167549, 0.042137693613767624, -0.0630110427737236, -0.006...
The Differentiable Cross-Entropy Method
https://proceedings.mlr.press/v119/amos20a.html
[ "Brandon Amos", "Denis Yarats" ]
null
null
We study the Cross-Entropy Method (CEM) for the non-convex optimization of a continuous and parameterized objective function and introduce a differentiable variant that enables us to differentiate the output of CEM with respect to the objective function’s parameters. In the machine learning setting this brings CEM insi...
[]
null
28
1909.12830
title_snapshot
[ -0.02461753413081169, 0.0039125774055719376, -0.01868625357747078, -0.0015102088218554854, 0.042483653873205185, 0.04069626331329346, 0.018311407417058945, -0.02349134534597397, -0.010092892684042454, -0.04778140038251877, 0.005730282049626112, 0.005249705631285906, -0.05307898297905922, -...
Customizing ML Predictions for Online Algorithms
https://proceedings.mlr.press/v119/anand20a.html
[ "Keerti Anand", "Rong Ge", "Debmalya Panigrahi" ]
null
null
A popular line of recent research incorporates ML advice in the design of online algorithms to improve their performance in typical instances. These papers treat the ML algorithm as a black-box, and redesign online algorithms to take advantage of ML predictions. In this paper, we ask the complementary question: can we ...
[]
null
29
2205.08715
title_snapshot
[ -0.0081414096057415, -0.014692392200231552, 0.0041344910860061646, 0.025502311065793037, 0.05939408391714096, 0.03329239785671234, 0.015753481537103653, -0.006270741578191519, -0.016360895708203316, -0.023257313296198845, -0.018603269010782242, 0.004446056205779314, -0.067036472260952, -0....
Fairwashing explanations with off-manifold detergent
https://proceedings.mlr.press/v119/anders20a.html
[ "Christopher Anders", "Plamen Pasliev", "Ann-Kathrin Dombrowski", "Klaus-Robert Müller", "Pan Kessel" ]
null
null
Explanation methods promise to make black-box classifiers more transparent. As a result, it is hoped that they can act as proof for a sensible, fair and trustworthy decision-making process of the algorithm and thereby increase its acceptance by the end-users. In this paper, we show both theoretically and experimentally...
[]
null
30
2007.09969
title_snapshot
[ -0.002016089390963316, -0.03850831091403961, -0.021623067557811737, 0.043462660163640976, 0.0007464094669558108, 0.027471037581562996, 0.02014852873980999, 0.0003402371658012271, -0.02897648885846138, -0.04054904356598854, -0.011575095355510712, 0.02078866958618164, -0.088096022605896, 0.0...
Population-Based Black-Box Optimization for Biological Sequence Design
https://proceedings.mlr.press/v119/angermueller20a.html
[ "Christof Angermueller", "David Belanger", "Andreea Gane", "Zelda Mariet", "David Dohan", "Kevin Murphy", "Lucy Colwell", "D Sculley" ]
null
null
The use of black-box optimization for the design of new biological sequences is an emerging research area with potentially revolutionary impact. The cost and latency of wet-lab experiments requires methods that find good sequences in few experimental rounds of large batches of sequences — a setting that off-the-shelf b...
[]
null
31
2006.03227
title_snapshot
[ -0.016557609662413597, -0.015682246536016464, -0.020407702773809433, 0.0314694382250309, 0.04369866102933884, 0.03763944283127785, 0.018021129071712494, 0.0045102545991539955, 0.016104578971862793, -0.040025606751441956, 0.01965535432100296, 0.01182324904948473, -0.07815235108137131, -0.01...
Low-loss connection of weight vectors: distribution-based approaches
https://proceedings.mlr.press/v119/anokhin20a.html
[ "Ivan Anokhin", "Dmitry Yarotsky" ]
null
null
Recent research shows that sublevel sets of the loss surfaces of overparameterized networks are connected, exactly or approximately. We describe and compare experimentally a panel of methods used to connect two low-loss points by a low-loss curve on this surface. Our methods vary in accuracy and complexity. Most of our...
[]
null
32
2008.00741
title_snapshot
[ -0.024491216987371445, -0.010432140901684761, 0.008245501667261124, 0.02158127725124359, 0.036951325833797455, 0.033622074872255325, 0.006686476059257984, 0.005086076445877552, -0.005878749303519726, -0.055340271443128586, 0.013601528480648994, -0.008763734251260757, -0.06873341649770737, ...
Online metric algorithms with untrusted predictions
https://proceedings.mlr.press/v119/antoniadis20a.html
[ "Antonios Antoniadis", "Christian Coester", "Marek Elias", "Adam Polak", "Bertrand Simon" ]
null
null
Machine-learned predictors, although achieving very good results for inputs resembling training data, cannot possibly provide perfect predictions in all situations. Still, decision-making systems that are based on such predictors need not only to benefit from good predictions but also to achieve a decent performance wh...
[]
null
33
2003.02144
title_snapshot
[ -0.019426973536610603, 0.0102030448615551, -0.022162102162837982, 0.049351610243320465, 0.05933938920497894, 0.015261435881257057, 0.020105969160795212, 0.03530609980225563, -0.013234303332865238, -0.022451354190707207, -0.027129974216222763, 0.010587755590677261, -0.08607064932584763, -0....
NADS: Neural Architecture Distribution Search for Uncertainty Awareness
https://proceedings.mlr.press/v119/ardywibowo20a.html
[ "Randy Ardywibowo", "Shahin Boluki", "Xinyu Gong", "Zhangyang Wang", "Xiaoning Qian" ]
null
null
Machine learning (ML) systems often encounter Out-of-Distribution (OoD) errors when dealing with testing data coming from a distribution different from training data. It becomes important for ML systems in critical applications to accurately quantify its predictive uncertainty and screen out these anomalous inputs. How...
[]
null
34
2006.06646
title_snapshot
[ -0.01509853359311819, 0.01050485298037529, -0.043453674763441086, 0.048739880323410034, 0.03771571069955826, 0.023483525961637497, 0.007193397264927626, -0.00654206657782197, -0.03313567861914635, -0.044136516749858856, -0.010647958144545555, -0.003137102583423257, -0.037120040506124496, -...
Provable Representation Learning for Imitation Learning via Bi-level Optimization
https://proceedings.mlr.press/v119/arora20a.html
[ "Sanjeev Arora", "Simon Du", "Sham Kakade", "Yuping Luo", "Nikunj Saunshi" ]
null
null
A common strategy in modern learning systems is to learn a representation that is useful for many tasks, a.k.a. representation learning. We study this strategy in the imitation learning setting for Markov decision processes (MDPs) where multiple experts’ trajectories are available. We formulate representation learning ...
[]
null
35
2002.10544
title_snapshot
[ -0.043846845626831055, -0.012314921244978905, -0.017543263733386993, 0.03794170171022415, 0.046304117888212204, 0.029902860522270203, 0.03296063840389252, 0.006397385150194168, -0.03043168969452381, -0.022198395803570747, 0.003863590070977807, -0.010115290060639381, -0.06396261602640152, -...
Quantum Boosting
https://proceedings.mlr.press/v119/arunachalam20a.html
[ "Srinivasan Arunachalam", "Reevu Maity" ]
null
null
Boosting is a technique that boosts a weak and inaccurate machine learning algorithm into a strong accurate learning algorithm. The AdaBoost algorithm by Freund and Schapire (for which they were awarded the G{ö}del prize in 2003) is one of the widely used boosting algorithms, with many applications in theory and practi...
[]
null
36
2002.05056
title_snapshot
[ -0.03485051169991493, -0.02674020268023014, -0.020696179941296577, 0.026928117498755455, 0.030263977125287056, -0.023344656452536583, 0.024710407480597496, -0.018770897760987282, -0.016736412420868874, -0.040408696979284286, -0.04088018089532852, 0.0016661783447489142, -0.08296419680118561, ...
Black-box Certification and Learning under Adversarial Perturbations
https://proceedings.mlr.press/v119/ashtiani20a.html
[ "Hassan Ashtiani", "Vinayak Pathak", "Ruth Urner" ]
null
null
We formally study the problem of classification under adversarial perturbations from a learner’s perspective as well as a third-party who aims at certifying the robustness of a given black-box classifier. We analyze a PAC-type framework of semi-supervised learning and identify possibility and impossibility results for ...
[]
null
37
2006.16520
title_snapshot
[ -0.0071055092848837376, -0.025272369384765625, -0.03205745294690132, 0.052032142877578735, 0.013955159112811089, 0.012205072678625584, 0.034626420587301254, -0.01811559870839119, -0.02802988700568676, -0.017234941944479942, -0.034871432930231094, 0.019644320011138916, -0.06191883608698845, ...
Invertible generative models for inverse problems: mitigating representation error and dataset bias
https://proceedings.mlr.press/v119/asim20a.html
[ "Muhammad Asim", "Mara Daniels", "Oscar Leong", "Ali Ahmed", "Paul Hand" ]
null
null
Trained generative models have shown remarkable performance as priors for inverse problems in imaging – for example, Generative Adversarial Network priors permit recovery of test images from 5-10x fewer measurements than sparsity priors. Unfortunately, these models may be unable to represent any particular image becaus...
[]
null
38
1905.11672
title_snapshot
[ 0.001621495932340622, -0.01609453186392784, -0.033853061497211456, 0.04961830750107765, 0.046333614736795425, 0.030816366896033287, 0.03721628710627556, 0.00006776380905648693, -0.03224797919392586, -0.07933359593153, 0.004975828342139721, -0.012130418792366982, -0.028240999206900597, -0.0...
On the Convergence of Nesterov’s Accelerated Gradient Method in Stochastic Settings
https://proceedings.mlr.press/v119/assran20a.html
[ "Mahmoud Assran", "Mike Rabbat" ]
null
null
We study Nesterov’s accelerated gradient method with constant step-size and momentum parameters in the stochastic approximation setting (unbiased gradients with bounded variance) and the finite-sum setting (where randomness is due to sampling mini-batches). To build better insight into the behavior of Nesterov’s method...
[]
null
39
2002.12414
title_snapshot
[ -0.041464366018772125, -0.010487384162843227, 0.0023706972133368254, 0.014367274940013885, 0.022872375324368477, 0.049252867698669434, 0.04462924227118492, 0.02639639936387539, -0.028084000572562218, -0.05846099555492401, 0.03002127818763256, -0.03771241381764412, -0.055527567863464355, -0...
Safe screening rules for L0-regression from Perspective Relaxations
https://proceedings.mlr.press/v119/atamturk20a.html
[ "Alper Atamturk", "Andres Gomez" ]
null
null
We give safe screening rules to eliminate variables from regression with $\ell_0$ regularization or cardinality constraint. These rules are based on guarantees that a feature may or may not be selected in an optimal solution. The screening rules can be computed from a convex relaxation solution in linear time, without ...
[]
null
40
2004.08773
title_judge
[ -0.03391525521874428, -0.005962214432656765, 0.0003755865909624845, 0.000881710322573781, 0.05379312485456467, 0.051045652478933334, 0.019117126241326332, -0.015761317685246468, -0.03763701021671295, -0.02898283675312996, -0.04032718390226364, 0.03555415943264961, -0.06195726990699768, 0.0...
Adversarial Learning Guarantees for Linear Hypotheses and Neural Networks
https://proceedings.mlr.press/v119/awasthi20a.html
[ "Pranjal Awasthi", "Natalie Frank", "Mehryar Mohri" ]
null
null
Adversarial or test time robustness measures the susceptibility of a classifier to perturbations to the test input. While there has been a flurry of recent work on designing defenses against such perturbations, the theory of adversarial robustness is not well understood. In order to make progress on this, we focus on t...
[]
null
41
2004.13617
title_snapshot
[ -0.020507829263806343, 0.010226758196949959, 0.00346726574935019, 0.047742269933223724, 0.02493441477417946, 0.01216217316687107, 0.05013449862599373, -0.02401386760175228, -0.01926915906369686, -0.032747793942689896, -0.014924171380698681, -0.013539539650082588, -0.07671861350536346, -0.0...
Sample Amplification: Increasing Dataset Size even when Learning is Impossible
https://proceedings.mlr.press/v119/axelrod20a.html
[ "Brian Axelrod", "Shivam Garg", "Vatsal Sharan", "Gregory Valiant" ]
null
null
Given data drawn from an unknown distribution, D, to what extent is it possible to “amplify” this dataset and faithfully output an even larger set of samples that appear to have been drawn from D? We formalize this question as follows: an (n,m) amplification procedure takes as input n independent draws from an unknown ...
[]
null
42
1904.12053
title_snapshot
[ 0.004452775232493877, -0.0323924794793129, -0.027296971529722214, 0.051544636487960815, 0.048857688903808594, 0.040193699300289154, 0.042285654693841934, -0.016086332499980927, -0.03029838763177395, -0.04736888036131859, -0.0014680711319670081, -0.01365276426076889, -0.09081704914569855, 0...
Sparse Convex Optimization via Adaptively Regularized Hard Thresholding
https://proceedings.mlr.press/v119/axiotis20a.html
[ "Kyriakos Axiotis", "Maxim Sviridenko" ]
null
null
The goal of Sparse Convex Optimization is to optimize a convex function $f$ under a sparsity constraint $s\leq s^*\gamma$, where $s^*$ is the target number of non-zero entries in a feasible solution (sparsity) and $\gamma\geq 1$ is an approximation factor. There has been a lot of work to analyze the sparsity guarantees...
[]
null
43
2006.14571
title_snapshot
[ -0.039643656462430954, -0.007635867223143578, 0.023990662768483162, 0.011845871806144714, 0.04180198162794113, 0.05427084118127823, 0.025113819167017937, -0.0014568414771929383, -0.05250605195760727, -0.049950238317251205, -0.016746625304222107, -0.0017750852275639772, -0.035393714904785156,...
Model-Based Reinforcement Learning with Value-Targeted Regression
https://proceedings.mlr.press/v119/ayoub20a.html
[ "Alex Ayoub", "Zeyu Jia", "Csaba Szepesvari", "Mengdi Wang", "Lin Yang" ]
null
null
This paper studies model-based reinforcement learning (RL) for regret minimization. We focus on finite-horizon episodic RL where the transition model $P$ belongs to a known family of models $\mathcal{P}$, a special case of which is when models in $\mathcal{P}$ take the form of linear mixtures: $P_{\theta} = \sum_{i=1}^...
[]
null
44
2006.01107
title_snapshot
[ -0.03455955907702446, -0.010894080623984337, -0.008439815603196621, 0.02365606091916561, 0.04938764125108719, 0.04759897664189339, 0.016225989907979965, -0.003921966068446636, -0.03070639632642269, -0.032539959996938705, -0.03409484401345253, 0.028566032648086548, -0.06501714885234833, -0....
Forecasting Sequential Data Using Consistent Koopman Autoencoders
https://proceedings.mlr.press/v119/azencot20a.html
[ "Omri Azencot", "N. Benjamin Erichson", "Vanessa Lin", "Michael Mahoney" ]
null
null
Recurrent neural networks are widely used on time series data, yet such models often ignore the underlying physical structures in such sequences. A new class of physics-based methods related to Koopman theory has been introduced, offering an alternative for processing nonlinear dynamical systems. In this work, we propo...
[]
null
45
2003.02236
title_snapshot
[ -0.03871564939618111, -0.03495805710554123, -0.04833941534161568, 0.05109281465411186, 0.049789778888225555, 0.046032704412937164, 0.03638443350791931, 0.00794405397027731, -0.0476430244743824, -0.029449649155139923, 0.014823336154222488, 0.014585223980247974, -0.07759717851877213, 0.01769...
Constant Curvature Graph Convolutional Networks
https://proceedings.mlr.press/v119/bachmann20a.html
[ "Gregor Bachmann", "Gary Becigneul", "Octavian Ganea" ]
null
null
Interest has been rising lately towards methods representing data in non-Euclidean spaces, e.g. hyperbolic or spherical that provide specific inductive biases useful for certain real-world data properties, e.g. scale-free, hierarchical or cyclical. However, the popular graph neural networks are currently limited in mod...
[]
null
46
1911.05076
title_snapshot
[ -0.0003609709965530783, -0.009177874773740768, 0.022984500974416733, 0.04578391835093498, 0.013466974720358849, 0.025878002867102623, 0.018807807937264442, 0.04453658312559128, -0.026698224246501923, -0.07053320854902267, -0.007481482345610857, -0.026013771072030067, -0.04849863424897194, ...
Scalable Nearest Neighbor Search for Optimal Transport
https://proceedings.mlr.press/v119/backurs20a.html
[ "Arturs Backurs", "Yihe Dong", "Piotr Indyk", "Ilya Razenshteyn", "Tal Wagner" ]
null
null
The Optimal Transport (a.k.a. Wasserstein) distance is an increasingly popular similarity measure for rich data domains, such as images or text documents. This raises the necessity for fast nearest neighbor search algorithms according to this distance, which poses a substantial computational bottleneck on massive datas...
[]
null
47
1910.04126
title_snapshot
[ -0.047831322997808456, -0.048232682049274445, 0.030148735269904137, 0.04319797456264496, 0.028677914291620255, 0.02381817251443863, 0.019389238208532333, -0.0029260185547173023, -0.004649253562092781, -0.06738811731338501, -0.02045365236699581, -0.038313038647174835, -0.05637923255562782, ...
Agent57: Outperforming the Atari Human Benchmark
https://proceedings.mlr.press/v119/badia20a.html
[ "Adrià Puigdomènech Badia", "Bilal Piot", "Steven Kapturowski", "Pablo Sprechmann", "Alex Vitvitskyi", "Zhaohan Daniel Guo", "Charles Blundell" ]
null
null
Atari games have been a long-standing benchmark in the reinforcement learning (RL) community for the past decade. This benchmark was proposed to test general competency of RL algorithms. Previous work has achieved good average performance by doing outstandingly well on many games of the set, but very poorly in several ...
[]
null
48
2003.13350
title_snapshot
[ -0.04743754491209984, -0.0253453329205513, -0.020331887528300285, 0.03413916006684303, 0.060261718928813934, 0.020001105964183807, 0.016495373100042343, 0.005244252271950245, -0.03816549852490425, -0.04637051746249199, 0.0006308801239356399, 0.03592563420534134, -0.05512293800711632, -0.01...
Fiduciary Bandits
https://proceedings.mlr.press/v119/bahar20a.html
[ "Gal Bahar", "Omer Ben-Porat", "Kevin Leyton-Brown", "Moshe Tennenholtz" ]
null
null
Recommendation systems often face exploration-exploitation tradeoffs: the system can only learn about the desirability of new options by recommending them to some user. Such systems can thus be modeled as multi-armed bandit settings; however, users are self-interested and cannot be made to follow recommendations. We as...
[]
null
49
1905.07043
title_snapshot
[ 0.0036251796409487724, -0.03256797045469284, 0.01881948858499527, 0.04165443032979965, 0.058701153844594955, -0.0032719445880502462, 0.020326806232333183, 0.01386469230055809, 0.0031484446953982115, -0.027897167950868607, -0.011953773908317089, 0.022682588547468185, -0.04291956126689911, -...
Learning De-biased Representations with Biased Representations
https://proceedings.mlr.press/v119/bahng20a.html
[ "Hyojin Bahng", "Sanghyuk Chun", "Sangdoo Yun", "Jaegul Choo", "Seong Joon Oh" ]
null
null
Many machine learning algorithms are trained and evaluated by splitting data from a single source into training and test sets. While such focus on in-distribution learning scenarios has led to interesting advancement, it has not been able to tell if models are relying on dataset biases as shortcuts for successful predi...
[]
null
50
1910.02806
title_snapshot
[ 0.02314063534140587, -0.012221667915582657, -0.028626469895243645, 0.053672533482313156, 0.024181663990020752, 0.021358275786042213, 0.025805315002799034, 0.01040472649037838, -0.018972385674715042, -0.03650454431772232, -0.02858218550682068, -0.009864107705652714, -0.08993451297283173, 0....
Deep k-NN for Noisy Labels
https://proceedings.mlr.press/v119/bahri20a.html
[ "Dara Bahri", "Heinrich Jiang", "Maya Gupta" ]
null
null
Modern machine learning models are often trained on examples with noisy labels that hurt performance and are hard to identify. In this paper, we provide an empirical study showing that a simple $k$-nearest neighbor-based filtering approach on the logit layer of a preliminary model can remove mislabeled training data an...
[]
null
51
2004.12289
title_snapshot
[ -0.005785962101072073, -0.020655490458011627, -0.014282588846981525, 0.051582686603069305, 0.03897199034690857, 0.037243008613586426, 0.0121597396209836, -0.03127056360244751, -0.010831945575773716, -0.043898217380046844, -0.007739303633570671, 0.0014482996193692088, -0.07258284837007523, ...
Provable Self-Play Algorithms for Competitive Reinforcement Learning
https://proceedings.mlr.press/v119/bai20a.html
[ "Yu Bai", "Chi Jin" ]
null
null
Self-play, where the algorithm learns by playing against itself without requiring any direct supervision, has become the new weapon in modern Reinforcement Learning (RL) for achieving superhuman performance in practice. However, the majority of exisiting theory in reinforcement learning only applies to the setting wher...
[]
null
52
2002.04017
title_snapshot
[ -0.054609909653663635, -0.03900495171546936, -0.0070299203507602215, 0.026153912767767906, 0.057189296931028366, -0.001344872172921896, 0.02878877893090248, -0.00212834938429296, -0.01881079003214836, -0.021025437861680984, -0.020077109336853027, 0.0013980354415252805, -0.04967271536588669, ...
Sparse Subspace Clustering with Entropy-Norm
https://proceedings.mlr.press/v119/bai20b.html
[ "Liang Bai", "Jiye Liang" ]
null
null
In this paper, we provide an explicit theoretical connection between Sparse subspace clustering (SSC) and spectral clustering (SC) from the perspective of learning a data similarity matrix. We show that spectral clustering with Gaussian kernel can be viewed as sparse subspace clustering with entropy-norm (SSC+E). Compa...
[]
null
53
null
null
[ -0.0033418810926377773, -0.025225847959518433, 0.021793918684124947, 0.03360894322395325, 0.03224203363060951, 0.027435140684247017, 0.021528221666812897, -0.019010666757822037, -0.039574988186359406, -0.037867505103349686, -0.04473722726106644, -0.0009478762513026595, -0.06659902632236481, ...
Coresets for Clustering in Graphs of Bounded Treewidth
https://proceedings.mlr.press/v119/baker20a.html
[ "Daniel Baker", "Vladimir Braverman", "Lingxiao Huang", "Shaofeng H.-C. Jiang", "Robert Krauthgamer", "Xuan Wu" ]
null
null
We initiate the study of coresets for clustering in graph metrics, i.e., the shortest-path metric of edge-weighted graphs. Such clustering problems are essential to data analysis and used for example in road networks and data visualization. A coreset is a compact summary of the data that approximately preserves the clu...
[]
null
54
1907.04733
title_snapshot
[ -0.013482167385518551, -0.03434867039322853, 0.01688319258391857, 0.04341140761971474, 0.03171033784747124, 0.04255516454577446, 0.014769341796636581, 0.009693792089819908, -0.0022801379673182964, -0.04938645660877228, 0.00010792172543006018, -0.05551277846097946, -0.06781727075576782, -0....
Refined bounds for algorithm configuration: The knife-edge of dual class approximability
https://proceedings.mlr.press/v119/balcan20a.html
[ "Maria-Florina Balcan", "Tuomas Sandholm", "Ellen Vitercik" ]
null
null
Automating algorithm configuration is growing increasingly necessary as algorithms come with more and more tunable parameters. It is common to tune parameters using machine learning, optimizing algorithmic performance (runtime or solution quality, for example) using a training set of problem instances from the specific...
[]
null
55
2006.11827
title_snapshot
[ -0.046372171491384506, -0.00965997762978077, -0.02298976108431816, 0.032323677092790604, 0.05115698650479317, 0.03468719497323036, 0.03624214977025986, -0.037961795926094055, -0.019380295649170876, -0.017126886174082756, -0.013103128410875797, -0.01781352423131466, -0.08279912173748016, -0...
Ready Policy One: World Building Through Active Learning
https://proceedings.mlr.press/v119/ball20a.html
[ "Philip Ball", "Jack Parker-Holder", "Aldo Pacchiano", "Krzysztof Choromanski", "Stephen Roberts" ]
null
null
Model-Based Reinforcement Learning (MBRL) offers a promising direction for sample efficient learning, often achieving state of the art results for continuous control tasks. However many existing MBRL methods rely on combining greedy policies with exploration heuristics, and even those which utilize principled explorati...
[]
null
56
2002.02693
title_snapshot
[ -0.032035235315561295, -0.0198991559445858, -0.005944659933447838, 0.03818381205201149, 0.04577135667204857, 0.016568126156926155, -0.0018864767625927925, -0.011771230027079582, -0.021178212016820908, -0.029929736629128456, -0.01755470223724842, 0.019969122484326363, -0.09103215485811234, ...
Stochastic Optimization for Regularized Wasserstein Estimators
https://proceedings.mlr.press/v119/ballu20a.html
[ "Marin Ballu", "Quentin Berthet", "Francis Bach" ]
null
null
Optimal transport is a foundational problem in optimization, that allows to compare probability distributions while taking into account geometric aspects. Its optimal objective value, the Wasserstein distance, provides an important loss between distributions that has been used in many applications throughout machine le...
[]
null
57
2002.08695
title_snapshot
[ -0.03480900451540947, -0.01244070753455162, 0.022768186405301094, 0.021187705919146538, 0.03813473880290985, 0.029350552707910538, 0.013570568524301052, 0.019338233396410942, -0.01768055371940136, -0.06418053805828094, -0.014823910780251026, -0.010276599787175655, -0.0336311049759388, -0.0...
Dual Mirror Descent for Online Allocation Problems
https://proceedings.mlr.press/v119/balseiro20a.html
[ "Santiago Balseiro", "Haihao Lu", "Vahab Mirrokni" ]
null
null
We consider online allocation problems with concave revenue functions and resource constraints, which are central problems in revenue management and online advertising. In these settings, requests arrive sequentially during a finite horizon and, for each request, a decision maker needs to choose an action that consumes...
[]
null
58
2002.10421
title_snapshot
[ -0.032226499170064926, -0.012598522007465363, -0.009105653502047062, 0.026747344061732292, 0.04141372814774513, 0.051730457693338394, -0.00027754876646213233, 0.022256888449192047, -0.017322886735200882, -0.029119960963726044, -0.02148824743926525, 0.03513028472661972, -0.05205346271395683, ...
Inductive-bias-driven Reinforcement Learning For Efficient Schedules in Heterogeneous Clusters
https://proceedings.mlr.press/v119/banerjee20a.html
[ "Subho Banerjee", "Saurabh Jha", "Zbigniew Kalbarczyk", "Ravishankar Iyer" ]
null
null
The problem of scheduling of workloads onto heterogeneous processors (e.g., CPUs, GPUs, FPGAs) is of fundamental importance in modern data centers. Current system schedulers rely on application/system-specific heuristics that have to be built on a case-by-case basis. Recent work has demonstrated ML techniques for autom...
[]
null
59
1909.02119
title_snapshot
[ -0.01902650110423565, -0.03442981839179993, -0.018924947828054428, 0.06463344395160675, 0.034918226301670074, 0.011542883701622486, 0.019371846690773964, 0.01374098751693964, -0.03600160777568817, -0.03171424940228462, -0.01743011362850666, -0.0031135734170675278, -0.06739088147878647, 0.0...
UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training
https://proceedings.mlr.press/v119/bao20a.html
[ "Hangbo Bao", "Li Dong", "Furu Wei", "Wenhui Wang", "Nan Yang", "Xiaodong Liu", "Yu Wang", "Jianfeng Gao", "Songhao Piao", "Ming Zhou", "Hsiao-Wuen Hon" ]
null
null
We propose to pre-train a unified language model for both autoencoding and partially autoregressive language modeling tasks using a novel training procedure, referred to as a pseudo-masked language model (PMLM). Given an input text with masked tokens, we rely on conventional masks to learn inter-relations between corru...
[]
null
60
2002.12804
title_snapshot
[ 0.009157405234873295, -0.013689747080206871, -0.017780344933271408, 0.03914882615208626, 0.04292682185769081, 0.02351030893623829, 0.050469592213630676, 0.018537083640694618, -0.027926018461585045, -0.010543433018028736, -0.04075050354003906, 0.01756499521434307, -0.07130146026611328, 0.00...
Fast OSCAR and OWL Regression via Safe Screening Rules
https://proceedings.mlr.press/v119/bao20b.html
[ "Runxue Bao", "Bin Gu", "Heng Huang" ]
null
null
Ordered Weighted $L_{1}$ (OWL) regularized regression is a new regression analysis for high-dimensional sparse learning. Proximal gradient methods are used as standard approaches to solve OWL regression. However, it is still a burning issue to solve OWL regression due to considerable computational cost and memory usage...
[]
null
61
2006.16433
title_snapshot
[ -0.032042790204286575, -0.01871296390891075, 0.011236338876187801, 0.016225093975663185, 0.0643211379647255, 0.02798980474472046, 0.006902516819536686, -0.020071202889084816, -0.034743085503578186, -0.02023453824222088, -0.011043847538530827, 0.03423751890659332, -0.08073145896196365, -0.0...
Option Discovery in the Absence of Rewards with Manifold Analysis
https://proceedings.mlr.press/v119/bar20a.html
[ "Amitay Bar", "Ronen Talmon", "Ron Meir" ]
null
null
Options have been shown to be an effective tool in reinforcement learning, facilitating improved exploration and learning. In this paper, we present an approach based on spectral graph theory and derive an algorithm that systematically discovers options without access to a specific reward or task assignment. As opposed...
[]
null
62
2003.05878
title_snapshot
[ -0.06093638762831688, -0.042973387986421585, 0.003782035782933235, 0.049110736697912216, 0.0430992990732193, 0.029471473768353462, 0.011476735584437847, 0.0023565690498799086, -0.01732003130018711, -0.0824657678604126, -0.006528571713715792, 0.04622165486216545, -0.08283419907093048, -0.00...
Learning the piece-wise constant graph structure of a varying Ising model
https://proceedings.mlr.press/v119/bars20a.html
[ "Batiste Le Bars", "Pierre Humbert", "Argyris Kalogeratos", "Nicolas Vayatis" ]
null
null
This work focuses on the estimation of multiple change-points in a time-varying Ising model that evolves piece-wise constantly. The aim is to identify both the moments at which significant changes occur in the Ising model, as well as the underlying graph structures. For this purpose, we propose to estimate the neighbor...
[]
null
63
1910.08512
title_snapshot
[ -0.0012642531655728817, -0.029062658548355103, 0.004148269072175026, 0.013750291429460049, 0.053759269416332245, 0.021829955279827118, 0.04004941135644913, 0.022857563570141792, -0.027143768966197968, -0.0373387448489666, 0.0023206076584756374, -0.01597004197537899, -0.057685405015945435, ...
Frequency Bias in Neural Networks for Input of Non-Uniform Density
https://proceedings.mlr.press/v119/basri20a.html
[ "Ronen Basri", "Meirav Galun", "Amnon Geifman", "David Jacobs", "Yoni Kasten", "Shira Kritchman" ]
null
null
Recent works have partly attributed the generalization ability of over-parameterized neural networks to frequency bias – networks trained with gradient descent on data drawn from a uniform distribution find a low frequency fit before high frequency ones. As realistic training sets are not drawn from a uniform distribut...
[]
null
64
2003.04560
title_snapshot
[ -0.026910914108157158, -0.02695036493241787, 0.010152747854590416, 0.05451817810535431, 0.016579123213887215, 0.033542148768901825, 0.011251446790993214, 0.021858802065253258, -0.024886315688490868, -0.05190436169505119, 0.004699993412941694, 0.019196635112166405, -0.0700017660856247, 0.00...
Private Query Release Assisted by Public Data
https://proceedings.mlr.press/v119/bassily20a.html
[ "Raef Bassily", "Albert Cheu", "Shay Moran", "Aleksandar Nikolov", "Jonathan Ullman", "Steven Wu" ]
null
null
We study the problem of differentially private query release assisted by access to public data. In this problem, the goal is to answer a large class $\mathcal{H}$ of statistical queries with error no more than $\alpha$ using a combination of public and private samples. The algorithm is required to satisfy differential ...
[]
null
65
2004.10941
title_snapshot
[ -0.027506226673722267, 0.017696110531687737, 0.001590954838320613, 0.0631428137421608, 0.037062376737594604, 0.0019869538955390453, 0.05252302065491676, -0.03538234159350395, -0.018865389749407768, -0.010220364667475224, -0.036752067506313324, -0.024707596749067307, -0.0642431229352951, -0...
ECLIPSE: An Extreme-Scale Linear Program Solver for Web-Applications
https://proceedings.mlr.press/v119/basu20a.html
[ "Kinjal Basu", "Amol Ghoting", "Rahul Mazumder", "Yao Pan" ]
null
null
Key problems arising in web applications (with millions of users and thousands of items) can be formulated as linear programs involving billions to trillions of decision variables and constraints. Despite the appeal of linear program (LP) formulations, solving problems at these scales appear to be well beyond the capab...
[]
null
66
null
null
[ -0.025799207389354706, -0.022408146411180496, -0.016396058723330498, 0.020707041025161743, 0.024801529943943024, 0.030505632981657982, 0.018319008871912956, -0.0026926409918814898, -0.024909375235438347, -0.04764216020703316, -0.006368663161993027, -0.01813138648867607, -0.09109726548194885,...
On Second-Order Group Influence Functions for Black-Box Predictions
https://proceedings.mlr.press/v119/basu20b.html
[ "Samyadeep Basu", "Xuchen You", "Soheil Feizi" ]
null
null
With the rapid adoption of machine learning systems in sensitive applications, there is an increasing need to make black-box models explainable. Often we want to identify an influential group of training samples in a particular test prediction for a given machine learning model. Existing influence functions tackle this...
[]
null
67
1911.00418
title_snapshot
[ -0.022254779934883118, -0.008000455796718597, 0.007771777454763651, 0.014619647525250912, 0.031182942911982536, 0.02960134856402874, 0.012861748225986958, -0.01462160050868988, 0.0015600729966536164, -0.03926680609583855, -0.011165644973516464, 0.036900632083415985, -0.07172328233718872, -...
Kernel interpolation with continuous volume sampling
https://proceedings.mlr.press/v119/belhadji20a.html
[ "Ayoub Belhadji", "Rémi Bardenet", "Pierre Chainais" ]
null
null
A fundamental task in kernel methods is to pick nodes and weights, so as to approximate a given function from an RKHS by the weighted sum of kernel translates located at the nodes. This is the crux of kernel density estimation, kernel quadrature, or interpolation from discrete samples. Furthermore, RKHSs offer a conven...
[]
null
68
2002.09677
title_snapshot
[ -0.010811619460582733, -0.021441033110022545, 0.028981458395719528, 0.043507520109415054, 0.023216577246785164, 0.05802486464381218, -0.0011288602836430073, -0.03364730253815651, -0.02593902312219143, -0.06160431355237961, -0.016064003109931946, -0.028779281303286552, -0.053569842129945755, ...
Decoupled Greedy Learning of CNNs
https://proceedings.mlr.press/v119/belilovsky20a.html
[ "Eugene Belilovsky", "Michael Eickenberg", "Edouard Oyallon" ]
null
null
A commonly cited inefficiency of neural network training by back-propagation is the update locking problem: each layer must wait for the signal to propagate through the network before updating. In recent years multiple authors have considered alternatives that can alleviate this issue. In this context, we consider a si...
[]
null
69
1901.08164
title_snapshot
[ -0.0012381888227537274, -0.010173549875617027, -0.006043196190148592, 0.042967088520526886, 0.05412263423204422, 0.04230090230703354, -0.014315432868897915, 0.002151327906176448, -0.023965373635292053, -0.04891539365053177, -0.0032196654938161373, 0.005219665355980396, -0.08471450209617615, ...
The Cost-free Nature of Optimally Tuning Tikhonov Regularizers and Other Ordered Smoothers
https://proceedings.mlr.press/v119/bellec20a.html
[ "Pierre Bellec", "Dana Yang" ]
null
null
We consider the problem of selecting the best estimator among a family of Tikhonov regularized estimators, or, alternatively, to select a linear combination of these regularizers that is as good as the best regularizer in the family. Our theory reveals that if the Tikhonov regularizers share the same penalty matrix wit...
[]
null
70
1905.12517
title_snapshot
[ -0.04970299452543259, -0.008455402217805386, 0.028165677562355995, -0.010249411687254906, 0.049903687089681625, 0.04471854120492935, 0.03182455524802208, 0.012017262168228626, -0.025775592774152756, -0.05124834552407265, -0.029150474816560745, 0.02312551625072956, -0.0729696974158287, -0.0...
Defense Through Diverse Directions
https://proceedings.mlr.press/v119/bender20a.html
[ "Christopher Bender", "Yang Li", "Yifeng Shi", "Michael K. Reiter", "Junier Oliva" ]
null
null
In this work we develop a novel Bayesian neural network methodology to achieve strong adversarial robustness without the need for online adversarial training. Unlike previous efforts in this direction, we do not rely solely on the stochasticity of network weights by minimizing the divergence between the learned paramet...
[]
null
71
2003.10602
title_snapshot
[ -0.0036358823999762535, 0.005478729959577322, -0.02461051754653454, 0.04289673641324043, 0.014549301005899906, 0.02870580367743969, 0.0357506088912487, -0.033704064786434174, -0.025817032903432846, -0.0458151139318943, 0.003195070428773761, 0.002475625602528453, -0.056036293506622314, -0.0...
Interference and Generalization in Temporal Difference Learning
https://proceedings.mlr.press/v119/bengio20a.html
[ "Emmanuel Bengio", "Joelle Pineau", "Doina Precup" ]
null
null
We study the link between generalization and interference in temporal-difference (TD) learning. Interference is defined as the inner product of two different gradients, representing their alignment; this quantity emerges as being of interest from a variety of observations about neural networks, parameter sharing and th...
[]
null
72
2003.06350
title_snapshot
[ -0.017139730975031853, 0.00024041044525802135, -0.028584502637386322, 0.037259120494127274, 0.03371241316199303, 0.0037589604035019875, 0.049039579927921295, 0.008250282146036625, -0.0354062095284462, -0.016484463587403297, 0.008390320464968681, 0.023463943973183632, -0.0621924065053463, 0...
Preselection Bandits
https://proceedings.mlr.press/v119/bengs20a.html
[ "Viktor Bengs", "Eyke Hüllermeier" ]
null
null
In this paper, we introduce the Preselection Bandit problem, in which the learner preselects a subset of arms (choice alternatives) for a user, which then chooses the final arm from this subset. The learner is not aware of the user’s preferences, but can learn them from observed choices. In our concrete setting, we all...
[]
null
73
1907.06123
title_snapshot
[ -0.031109750270843506, -0.013078812509775162, 0.0027622964698821306, 0.05432632565498352, 0.0424015074968338, 0.02382085658609867, 0.004646679852157831, 0.026076730340719223, -0.024840064346790314, -0.0372793935239315, -0.028482407331466675, 0.02052895352244377, -0.05494674667716026, -0.06...
Efficient Policy Learning from Surrogate-Loss Classification Reductions
https://proceedings.mlr.press/v119/bennett20a.html
[ "Andrew Bennett", "Nathan Kallus" ]
null
null
Recent work on policy learning from observational data has highlighted the importance of efficient policy evaluation and has proposed reductions to weighted (cost-sensitive) classification. But, efficient policy evaluation need not yield efficient estimation of policy parameters. We consider the estimation problem give...
[]
null
74
2002.05153
title_snapshot
[ -0.029738739132881165, -0.043858788907527924, 0.003952913451939821, 0.03659692406654358, 0.012059594504535198, 0.04517831280827522, -0.005819987505674362, -0.04865153878927231, -0.006680166348814964, -0.044160909950733185, 0.004669548012316227, 0.005780303850769997, -0.06652715802192688, -...
Training Neural Networks for and by Interpolation
https://proceedings.mlr.press/v119/berrada20a.html
[ "Leonard Berrada", "Andrew Zisserman", "M. Pawan Kumar" ]
null
null
In modern supervised learning, many deep neural networks are able to interpolate the data: the empirical loss can be driven to near zero on all samples simultaneously. In this work, we explicitly exploit this interpolation property for the design of a new optimization algorithm for deep learning, which we term Adaptive...
[]
null
75
1906.05661
title_snapshot
[ 0.003940968308597803, -0.007178547326475382, 0.013691081665456295, 0.027786321938037872, 0.03733249008655548, 0.03202151134610176, 0.014891588129103184, 0.004050564952194691, 0.0036915410310029984, -0.05119970068335533, -0.009025315754115582, -0.01289201807230711, -0.05133552849292755, -0....
Implicit differentiation of Lasso-type models for hyperparameter optimization
https://proceedings.mlr.press/v119/bertrand20a.html
[ "Quentin Bertrand", "Quentin Klopfenstein", "Mathieu Blondel", "Samuel Vaiter", "Alexandre Gramfort", "Joseph Salmon" ]
null
null
Setting regularization parameters for Lasso-type estimators is notoriously difficult, though crucial for obtaining the best accuracy. The most popular hyperparameter optimization approach is grid-search on a held-out dataset. However, grid-search requires to choose a predefined grid of parameters and scales exponential...
[]
null
76
2002.08943
title_snapshot
[ -0.045612987130880356, -0.0045711686834692955, -0.0010633841156959534, 0.0181584395468235, 0.03722786158323288, 0.04530588537454605, 0.032310184091329575, -0.03352522477507591, -0.028580838814377785, -0.036052800714969635, 0.008378082886338234, -0.022633938118815422, -0.05205485224723816, ...
Online Learning with Imperfect Hints
https://proceedings.mlr.press/v119/bhaskara20a.html
[ "Aditya Bhaskara", "Ashok Cutkosky", "Ravi Kumar", "Manish Purohit" ]
null
null
We consider a variant of the classical online linear optimization problem in which at every step, the online player receives a “hint” vector before choosing the action for that round. Rather surprisingly, it was shown that if the hint vector is guaranteed to have a positive correlation with the cost vector, then the on...
[]
null
77
2002.04726
title_snapshot
[ -0.033761925995349884, -0.013807497918605804, -0.0009644987876527011, 0.05737786367535591, 0.02618006430566311, 0.03547224402427673, 0.011648658663034439, 0.012095319107174873, -0.012439454905688763, -0.02691841684281826, -0.02009037882089615, -0.0026550667826086283, -0.0682111531496048, -...
When are Non-Parametric Methods Robust?
https://proceedings.mlr.press/v119/bhattacharjee20a.html
[ "Robi Bhattacharjee", "Kamalika Chaudhuri" ]
null
null
A growing body of research has shown that many classifiers are susceptible to adversarial examples – small strategic modifications to test inputs that lead to misclassification. In this work, we study general non-parametric methods, with a view towards understanding when they are robust to these modifications. We estab...
[]
null
78
2003.06121
title_snapshot
[ 0.0006558013847097754, -0.03767615556716919, -0.003900074865669012, 0.06764409691095352, 0.021314015612006187, 0.053055666387081146, 0.028673147782683372, -0.037574850022792816, -0.017667170614004135, -0.060542769730091095, -0.01239191833883524, 0.006118748337030411, -0.07792533934116364, ...
Learning and Sampling of Atomic Interventions from Observations
https://proceedings.mlr.press/v119/bhattacharyya20a.html
[ "Arnab Bhattacharyya", "Sutanu Gayen", "Saravanan Kandasamy", "Ashwin Maran", "Vinodchandran N. Variyam" ]
null
null
We study the problem of efficiently estimating the effect of an intervention on a single variable using observational samples. Our goal is to give algorithms with polynomial time and sample complexity in a non-parametric setting. Tian and Pearl (AAAI ’02) have exactly characterized the class of causal graphs for which ...
[]
null
79
2002.04232
title_snapshot
[ -0.012860528193414211, 0.0043624634854495525, -0.03134296461939812, 0.035545095801353455, 0.03781486302614212, 0.024404918774962425, 0.028278613463044167, 0.004402501974254847, -0.0021820636466145515, -0.047110483050346375, 0.0029029445722699165, -0.005935887806117535, -0.08417617529630661, ...
Near-optimal sample complexity bounds for learning Latent $k-$polytopes and applications to Ad-Mixtures
https://proceedings.mlr.press/v119/bhattacharyya20b.html
[ "Chiranjib Bhattacharyya", "Ravindran Kannan" ]
null
null
Deriving Optimal bounds on Sample Complexity of Latent Variable models is an active area of research. Recently such bounds were obtained for Mixture of Gaussians \cite{HSNCAY18}, no such results are known for Ad-mixtures, a generalization of Mixture distributions. In this paper we show that $O^*(dk/m)$ samples are suff...
[]
null
80
null
null
[ -0.013493875972926617, -0.004974215757101774, 0.002822674810886383, 0.057042937725782394, 0.028382739052176476, 0.027613122016191483, 0.03610566258430481, -0.002728719962760806, -0.02974921464920044, -0.015604314394295216, -0.02217334881424904, -0.01374927256256342, -0.06701067090034485, 0...
Low-Rank Bottleneck in Multi-head Attention Models
https://proceedings.mlr.press/v119/bhojanapalli20a.html
[ "Srinadh Bhojanapalli", "Chulhee Yun", "Ankit Singh Rawat", "Sashank Reddi", "Sanjiv Kumar" ]
null
null
Attention based Transformer architecture has enabled significant advances in the field of natural language processing. In addition to new pre-training techniques, recent improvements crucially rely on working with a relatively larger embedding dimension for tokens. Unfortunately, this leads to models that are prohibiti...
[]
null
81
2002.07028
title_snapshot
[ -0.016885532066226006, -0.02231941558420658, 0.0178949236869812, 0.0372701920568943, -0.011242794804275036, 0.032772086560726166, 0.03438257426023483, 0.005371444392949343, -0.021253414452075958, -0.00008020974928513169, -0.022189101204276085, 0.01982560195028782, -0.059559714049100876, 0....
Spectral Clustering with Graph Neural Networks for Graph Pooling
https://proceedings.mlr.press/v119/bianchi20a.html
[ "Filippo Maria Bianchi", "Daniele Grattarola", "Cesare Alippi" ]
null
null
Spectral clustering (SC) is a popular clustering technique to find strongly connected communities on a graph. SC can be used in Graph Neural Networks (GNNs) to implement pooling operations that aggregate nodes belonging to the same cluster. However, the eigendecomposition of the Laplacian is expensive and, since cluste...
[]
null
82
1907.00481
title_snapshot
[ -0.00316237541846931, -0.04965418577194214, 0.017377210780978203, 0.05298906937241554, 0.03395237773656845, 0.04698564484715462, 0.017413679510354996, 0.0032842184882611036, -0.008818241767585278, -0.05343359336256981, 0.003954725340008736, -0.017374180257320404, -0.07359518855810165, 0.00...
Time Series Deconfounder: Estimating Treatment Effects over Time in the Presence of Hidden Confounders
https://proceedings.mlr.press/v119/bica20a.html
[ "Ioana Bica", "Ahmed Alaa", "Mihaela Van Der Schaar" ]
null
null
The estimation of treatment effects is a pervasive problem in medicine. Existing methods for estimating treatment effects from longitudinal observational data assume that there are no hidden confounders, an assumption that is not testable in practice and, if it does not hold, leads to biased estimates. In this paper, w...
[]
null
83
1902.00450
title_snapshot
[ 0.01922684721648693, -0.007648266851902008, -0.046813543885946274, 0.015659542754292488, 0.04212581738829613, 0.059428904205560684, 0.06157749518752098, -0.00017856512567959726, -0.007343268487602472, -0.04414035752415657, 0.029404882341623306, 0.005212090443819761, -0.05192306637763977, 0...
Adversarial Robustness for Code
https://proceedings.mlr.press/v119/bielik20a.html
[ "Pavol Bielik", "Martin Vechev" ]
null
null
Machine learning and deep learning in particular has been recently used to successfully address many tasks in the domain of code such as finding and fixing bugs, code completion, decompilation, type inference and many others. However, the issue of adversarial robustness of models for code has gone largely unnoticed. In...
[]
null
84
2002.04694
title_snapshot
[ 0.016076823696494102, -0.02191682532429695, -0.04526960104703903, 0.0500863716006279, 0.04206555336713791, 0.0235286895185709, 0.05093774199485779, -0.027262499555945396, -0.017246119678020477, -0.06342603266239166, -0.030802173539996147, -0.008792759850621223, -0.07628875225782394, 0.0101...
The Boomerang Sampler
https://proceedings.mlr.press/v119/bierkens20a.html
[ "Joris Bierkens", "Sebastiano Grazzi", "Kengo Kamatani", "Gareth Roberts" ]
null
null
This paper introduces the boomerang sampler as a novel class of continuous-time non-reversible Markov chain Monte Carlo algorithms. The methodology begins by representing the target density as a density, $e^{-U}$, with respect to a prescribed (usually) Gaussian measure and constructs a continuous trajectory consisting ...
[]
null
85
2006.13777
title_snapshot
[ -0.0159965381026268, -0.011112268082797527, 0.00019452249398455024, 0.03890473395586014, 0.04858607426285744, -0.008116690441966057, 0.019819151610136032, 0.0060033053159713745, -0.03954881802201271, -0.06138622760772705, 0.017071764916181564, -0.012492455542087555, -0.031023865565657616, ...
Tight Bounds on Minimax Regret under Logarithmic Loss via Self-Concordance
https://proceedings.mlr.press/v119/bilodeau20a.html
[ "Blair Bilodeau", "Dylan Foster", "Daniel Roy" ]
null
null
We consider the classical problem of sequential probability assignment under logarithmic loss while competing against an arbitrary, potentially nonparametric class of experts. We obtain tight bounds on the minimax regret via a new approach that exploits the self-concordance property of the logarithmic loss. We show tha...
[]
null
86
2007.01160
title_snapshot
[ -0.03530659154057503, -0.007786668371409178, -0.005275367293506861, 0.052246615290641785, 0.04789036512374878, 0.047437846660614014, 0.019974494352936745, 0.003810873255133629, -0.004990874323993921, -0.04360906034708023, -0.026146983727812767, 0.009839200414717197, -0.05641420930624008, -...
My Fair Bandit: Distributed Learning of Max-Min Fairness with Multi-player Bandits
https://proceedings.mlr.press/v119/bistritz20a.html
[ "Ilai Bistritz", "Tavor Baharav", "Amir Leshem", "Nicholas Bambos" ]
null
null
Consider N cooperative but non-communicating players where each plays one out of M arms for T turns. Players have different utilities for each arm, representable as an NxM matrix. These utilities are unknown to the players. In each turn players receive noisy observations of their utility for their selected arm. However...
[]
null
87
2002.09808
title_snapshot
[ -0.03423100709915161, -0.006794474553316832, -0.017349205911159515, 0.04183252528309822, 0.022883983328938484, 0.007208659313619137, 0.009909383952617645, 0.032153502106666565, -0.019923774525523186, -0.05875217914581299, 0.007987858727574348, 0.0032587270252406597, -0.06355409324169159, -...
Provable guarantees for decision tree induction: the agnostic setting
https://proceedings.mlr.press/v119/blanc20a.html
[ "Guy Blanc", "Jane Lange", "Li-Yang Tan" ]
null
null
We give strengthened provable guarantees on the performance of widely employed and empirically successful {\sl top-down decision tree learning heuristics}. While prior works have focused on the realizable setting, we consider the more realistic and challenging {\sl agnostic} setting. We show that for all monotone funct...
[]
null
88
2006.00743
title_snapshot
[ -0.02577689103782177, -0.019394753500819206, 0.002351403934881091, 0.04869040846824646, 0.036047596484422684, 0.027789892628788948, 0.03796129673719406, -0.03306280076503754, -0.019192541018128395, 0.010384917259216309, -0.030113445594906807, 0.02203787863254547, -0.07605993002653122, -0.0...
Fast Differentiable Sorting and Ranking
https://proceedings.mlr.press/v119/blondel20a.html
[ "Mathieu Blondel", "Olivier Teboul", "Quentin Berthet", "Josip Djolonga" ]
null
null
The sorting operation is one of the most commonly used building blocks in computer programming. In machine learning, it is often used for robust statistics. However, seen as a function, it is piecewise linear and as a result includes many kinks where it is non-differentiable. More problematic is the related ranking ope...
[]
null
89
2002.08871
title_snapshot
[ -0.044680047780275345, -0.011681783013045788, -0.002788821468129754, 0.008898970670998096, 0.03440214693546295, 0.025236323475837708, 0.007525307592004538, -0.012606477364897728, -0.039299529045820236, -0.04424146190285683, -0.0021359322126954794, -0.012685541063547134, -0.06520681828260422,...
Beyond Signal Propagation: Is Feature Diversity Necessary in Deep Neural Network Initialization?
https://proceedings.mlr.press/v119/blumenfeld20a.html
[ "Yaniv Blumenfeld", "Dar Gilboa", "Daniel Soudry" ]
null
null
Deep neural networks are typically initialized with random weights, with variances chosen to facilitate signal propagation and stable gradients. It is also believed that diversity of features is an important property of these initializations. We construct a deep convolutional network with identical features by initiali...
[]
null
90
2007.01038
title_snapshot
[ 0.014048295095562935, -0.025080690160393715, 0.0005626122583635151, 0.06035380810499191, 0.021672219038009644, 0.030821166932582855, 0.015014059841632843, 0.0018776609795168042, -0.019842443987727165, -0.05370477959513664, 0.000504460942465812, 0.0007466705865226686, -0.05644919350743294, ...
Modulating Surrogates for Bayesian Optimization
https://proceedings.mlr.press/v119/bodin20a.html
[ "Erik Bodin", "Markus Kaiser", "Ieva Kazlauskaite", "Zhenwen Dai", "Neill Campbell", "Carl Henrik Ek" ]
null
null
Bayesian optimization (BO) methods often rely on the assumption that the objective function is well-behaved, but in practice, this is seldom true for real-world objectives even if noise-free observations can be collected. Common approaches, which try to model the objective as precisely as possible, often fail to make p...
[]
null
91
1906.11152
title_snapshot
[ -0.016325663775205612, 0.006336239632219076, -0.006071501411497593, 0.028721211478114128, 0.03483935445547104, 0.04709242656826973, 0.020676376298069954, -0.010211281478404999, -0.005463087931275368, -0.04767216369509697, -0.008186504244804382, 0.013277645222842693, -0.0516616553068161, -0...
Deep Coordination Graphs
https://proceedings.mlr.press/v119/boehmer20a.html
[ "Wendelin Boehmer", "Vitaly Kurin", "Shimon Whiteson" ]
null
null
This paper introduces the deep coordination graph (DCG) for collaborative multi-agent reinforcement learning. DCG strikes a flexible trade-off between representational capacity and generalization by factoring the joint value function of all agents according to a coordination graph into payoffs between pairs of agents. ...
[]
null
92
1910.00091
title_snapshot
[ -0.02037118561565876, -0.02938620001077652, -0.015120980329811573, 0.055487606674432755, 0.03414934128522873, 0.006020074710249901, 0.015003112144768238, 0.011083168908953667, -0.024302272126078606, -0.04691179469227791, 0.028659190982580185, -0.0015963319456204772, -0.07476583123207092, -...
Lorentz Group Equivariant Neural Network for Particle Physics
https://proceedings.mlr.press/v119/bogatskiy20a.html
[ "Alexander Bogatskiy", "Brandon Anderson", "Jan Offermann", "Marwah Roussi", "David Miller", "Risi Kondor" ]
null
null
We present a neural network architecture that is fully equivariant with respect to transformations under the Lorentz group, a fundamental symmetry of space and time in physics. The architecture is based on the theory of the finite-dimensional representations of the Lorentz group and the equivariant nonlinearity involve...
[]
null
93
2006.04780
title_snapshot
[ -0.003927499055862427, 0.006220568437129259, 0.0035508903674781322, 0.02049235999584198, 0.00048632718971930444, 0.01888696663081646, 0.008477123454213142, 0.02687324397265911, -0.06298524141311646, -0.038323987275362015, -0.01823354698717594, -0.04324881359934807, -0.02866620197892189, 0....
Efficient Robustness Certificates for Discrete Data: Sparsity-Aware Randomized Smoothing for Graphs, Images and More
https://proceedings.mlr.press/v119/bojchevski20a.html
[ "Aleksandar Bojchevski", "Johannes Gasteiger", "Stephan Günnemann" ]
null
null
Existing techniques for certifying the robustness of models for discrete data either work only for a small class of models or are general at the expense of efficiency or tightness. Moreover, they do not account for sparsity in the input which, as our findings show, is often essential for obtaining non-trivial guarantee...
[]
null
94
2008.12952
title_snapshot
[ 0.001002472243271768, -0.02994086965918541, -0.0002677764859981835, 0.07102148234844208, 0.020706169307231903, 0.02820604108273983, 0.020526543259620667, 0.005523127969354391, -0.029391780495643616, -0.04641462117433548, -0.0026567764580249786, -0.034569889307022095, -0.06839945167303085, ...
Proper Network Interpretability Helps Adversarial Robustness in Classification
https://proceedings.mlr.press/v119/boopathy20a.html
[ "Akhilan Boopathy", "Sijia Liu", "Gaoyuan Zhang", "Cynthia Liu", "Pin-Yu Chen", "Shiyu Chang", "Luca Daniel" ]
null
null
Recent works have empirically shown that there exist adversarial examples that can be hidden from neural network interpretability (namely, making network interpretation maps visually similar), or interpretability is itself susceptible to adversarial attacks. In this paper, we theoretically show that with a proper measu...
[]
null
95
2006.14748
title_snapshot
[ -0.0062342239543795586, -0.0025114479940384626, -0.012196434661746025, 0.04184447601437569, 0.03854486346244812, 0.008701646700501442, 0.010137480683624744, -0.033698778599500656, -0.02969815582036972, -0.04028920829296112, -0.013765883632004261, -0.008644369430840015, -0.062062207609415054,...
Spectrum Dependent Learning Curves in Kernel Regression and Wide Neural Networks
https://proceedings.mlr.press/v119/bordelon20a.html
[ "Blake Bordelon", "Abdulkadir Canatar", "Cengiz Pehlevan" ]
null
null
We derive analytical expressions for the generalization performance of kernel regression as a function of the number of training samples using theoretical methods from Gaussian processes and statistical physics. Our expressions apply to wide neural networks due to an equivalence between training them and kernel regress...
[]
null
96
2002.02561
title_snapshot
[ -0.02571747824549675, -0.02911096252501011, 0.021980179473757744, 0.02936573326587677, 0.04456412047147751, 0.02499757707118988, 0.003203284228220582, -0.009782026521861553, -0.02810749039053917, -0.03336541727185249, -0.023332417011260986, 0.05670144036412239, -0.06098807230591774, 0.0202...
Small Data, Big Decisions: Model Selection in the Small-Data Regime
https://proceedings.mlr.press/v119/bornschein20a.html
[ "Jorg Bornschein", "Francesco Visin", "Simon Osindero" ]
null
null
Highly overparametrized neural networks can display curiously strong generalization performance – a phenomenon that has recently garnered a wealth of theoretical and empirical research in order to better understand it. In contrast to most previous work, which typically considers the performance as a function of the mod...
[]
null
97
2009.12583
title_snapshot
[ -0.06635075807571411, -0.028769323602318764, -0.0010576106142252684, 0.059512779116630554, 0.047976959496736526, 0.03512372449040413, 0.02037678472697735, 0.022979190573096275, -0.03045937791466713, -0.019778193905949593, -0.016453497111797333, 0.019659413024783134, -0.06993041187524796, -...
Latent Variable Modelling with Hyperbolic Normalizing Flows
https://proceedings.mlr.press/v119/bose20a.html
[ "Joey Bose", "Ariella Smofsky", "Renjie Liao", "Prakash Panangaden", "Will Hamilton" ]
null
null
The choice of approximate posterior distributions plays a central role in stochastic variational inference (SVI). One effective solution is the use of normalizing flows \cut{defined on Euclidean spaces} to construct flexible posterior distributions. However, one key limitation of existing normalizing flows is that they...
[]
null
98
2002.06336
title_snapshot
[ 0.0036816694773733616, -0.009195045568048954, 0.01693222112953663, 0.05048392713069916, 0.032841917127370834, 0.05113166943192482, 0.02914769947528839, -0.0006662347004748881, -0.01726522669196129, -0.07132122665643692, -0.00859603751450777, -0.00889052264392376, -0.04585179686546326, 0.02...
Tightening Exploration in Upper Confidence Reinforcement Learning
https://proceedings.mlr.press/v119/bourel20a.html
[ "Hippolyte Bourel", "Odalric Maillard", "Mohammad Sadegh Talebi" ]
null
null
The upper confidence reinforcement learning (UCRL2) algorithm introduced in \citep{jaksch2010near} is a popular method to perform regret minimization in unknown discrete Markov Decision Processes under the average-reward criterion. Despite its nice and generic theoretical regret guarantees, this algorithm and its varia...
[]
null
99
2004.09656
title_snapshot
[ -0.05018826946616173, -0.00307230232283473, -0.002948377514258027, 0.05347449332475662, 0.049939513206481934, 0.016225652769207954, 0.024302339181303978, 0.0023736997973173857, -0.024635400623083115, -0.021898550912737846, -0.02985587902367115, 0.01014742162078619, -0.06059950590133667, -0...
Preference Modeling with Context-Dependent Salient Features
https://proceedings.mlr.press/v119/bower20a.html
[ "Amanda Bower", "Laura Balzano" ]
null
null
We consider the problem of estimating a ranking on a set of items from noisy pairwise comparisons given item features. We address the fact that pairwise comparison data often reflects irrational choice, e.g. intransitivity. Our key observation is that two items compared in isolation from other items may be compared bas...
[]
null
100
2002.09615
title_snapshot
[ -0.015494155697524548, -0.0041624028235673904, -0.005299018695950508, 0.043380334973335266, 0.0175813976675272, 0.04299986734986305, 0.010580852627754211, 0.019218116998672485, -0.02832496538758278, -0.03293532505631447, -0.018598465248942375, 0.02972077764570713, -0.06332848966121674, -0....