title
stringlengths
13
137
paper_url
stringlengths
44
58
authors
listlengths
1
12
type
stringclasses
0 values
primary_area
stringclasses
0 values
abstract
large_stringlengths
331
1.97k
keywords
listlengths
0
0
TL;DR
large_stringclasses
0 values
submission_number
int64
1
434
arxiv_id
stringlengths
10
10
arxiv_id_source
stringclasses
2 values
embedding
listlengths
768
768
Uncovering Causality from Multivariate Hawkes Integrated Cumulants
https://proceedings.mlr.press/v70/achab17a.html
[ "Massil Achab", "Emmanuel Bacry", "Stéphane Gaı̈ffas", "Iacopo Mastromatteo", "Jean-François Muzy" ]
null
null
We design a new nonparametric method that allows one to estimate the matrix of integrated kernels of a multivariate Hawkes process. This matrix not only encodes the mutual influences of each node of the process, but also disentangles the causality relationships between them. Our approach is the first that leads to an e...
[]
null
1
1607.06333
title_snapshot
[ 0.01018630899488926, -0.03404493257403374, -0.015217969194054604, 0.0032867533154785633, 0.02615903504192829, 0.06361166387796402, 0.01702572964131832, 0.029018184170126915, 0.0038954184856265783, -0.039183132350444794, 0.02814246155321598, 0.009068401530385017, -0.025066856294870377, 0.01...
A Unified Maximum Likelihood Approach for Estimating Symmetric Properties of Discrete Distributions
https://proceedings.mlr.press/v70/acharya17a.html
[ "Jayadev Acharya", "Hirakendu Das", "Alon Orlitsky", "Ananda Theertha Suresh" ]
null
null
Symmetric distribution properties such as support size, support coverage, entropy, and proximity to uniformity, arise in many applications. Recently, researchers applied different estimators and analysis tools to derive asymptotically sample-optimal approximations for each of these properties. We show that a single, si...
[]
null
2
null
null
[ -0.0106377387419343, -0.008971471339464188, -0.0030553648248314857, 0.007529672235250473, 0.049125004559755325, 0.03558839112520218, 0.02216549776494503, 0.007811390794813633, -0.042793210595846176, -0.06151041015982628, 0.027910050004720688, -0.018035000190138817, -0.07734675705432892, -0...
Constrained Policy Optimization
https://proceedings.mlr.press/v70/achiam17a.html
[ "Joshua Achiam", "David Held", "Aviv Tamar", "Pieter Abbeel" ]
null
null
For many applications of reinforcement learning it can be more convenient to specify both a reward function and constraints, rather than trying to design behavior through the reward function. For example, systems that physically interact with or around humans should satisfy safety constraints. Recent advances in policy...
[]
null
3
1705.10528
title_snapshot
[ -0.04275535047054291, -0.02794565260410309, -0.02230040915310383, 0.06339455395936966, 0.039166953414678574, 0.023784883320331573, 0.02010919712483883, -0.016812393441796303, -0.03592335060238838, -0.032154977321624756, -0.0296158529818058, 0.0014803506201133132, -0.0645257830619812, -0.03...
The Price of Differential Privacy for Online Learning
https://proceedings.mlr.press/v70/agarwal17a.html
[ "Naman Agarwal", "Karan Singh" ]
null
null
We design differentially private algorithms for the problem of online linear optimization in the full information and bandit settings with optimal $O(T^{0.5})$ regret bounds. In the full-information setting, our results demonstrate that $\epsilon$-differential privacy may be ensured for free – in particular, the regret...
[]
null
4
1701.07953
title_snapshot
[ -0.002057969570159912, 0.027612177655100822, 0.0028304944280534983, 0.06588821113109589, 0.04478258639574051, 0.030029866844415665, 0.050198592245578766, -0.0066750068217515945, 0.0032339890021830797, -0.02114546112716198, -0.0022255864460021257, -0.018105655908584595, -0.06029149144887924, ...
Local Bayesian Optimization of Motor Skills
https://proceedings.mlr.press/v70/akrour17a.html
[ "Riad Akrour", "Dmitry Sorokin", "Jan Peters", "Gerhard Neumann" ]
null
null
Bayesian optimization is renowned for its sample efficiency but its application to higher dimensional tasks is impeded by its focus on global optimization. To scale to higher dimensional problems, we leverage the sample efficiency of Bayesian optimization in a local context. The optimization of the acquisition function...
[]
null
5
null
null
[ -0.024074193090200424, 0.02227005362510681, -0.0021131904795765877, 0.049727655947208405, 0.023051824420690536, 0.04494474455714226, 0.03682941570878029, -0.01446271687746048, -0.03865271434187889, -0.04476260393857956, 0.0016739668790251017, 0.008346429094672203, -0.04283944517374039, -0....
Connected Subgraph Detection with Mirror Descent on SDPs
https://proceedings.mlr.press/v70/aksoylar17a.html
[ "Cem Aksoylar", "Lorenzo Orecchia", "Venkatesh Saligrama" ]
null
null
We propose a novel, computationally efficient mirror-descent based optimization framework for subgraph detection in graph-structured data. Our aim is to discover anomalous patterns present in a connected subgraph of a given graph. This problem arises in many applications such as detection of network intrusions, communi...
[]
null
6
null
null
[ -0.006619173102080822, -0.027834167703986168, 0.0017937266966328025, 0.04821182042360306, 0.037636905908584595, 0.014141513034701347, 0.043083298951387405, -0.0032920031808316708, 0.00211209524422884, -0.049047596752643585, 0.018423212692141533, -0.0072503057308495045, -0.07203833758831024, ...
Learning from Clinical Judgments: Semi-Markov-Modulated Marked Hawkes Processes for Risk Prognosis
https://proceedings.mlr.press/v70/alaa17a.html
[ "Ahmed M. Alaa", "Scott Hu", "Mihaela Schaar" ]
null
null
Critically ill patients in regular wards are vulnerable to unanticipated adverse events which require prompt transfer to the intensive care unit (ICU). To allow for accurate prognosis of deteriorating patients, we develop a novel continuous-time probabilistic model for a monitored patient’s temporal sequence of physiol...
[]
null
7
1705.05267
title_snapshot
[ -0.0022869009990245104, 0.0032077166251838207, -0.04733778536319733, -0.005224280990660191, 0.06451328843832016, 0.025745727121829987, 0.04493672773241997, 0.021096128970384598, 0.0026225419715046883, -0.04711750149726868, 0.015290004201233387, 0.010804187506437302, -0.0337349995970726, 0....
A Semismooth Newton Method for Fast, Generic Convex Programming
https://proceedings.mlr.press/v70/ali17a.html
[ "Alnur Ali", "Eric Wong", "J. Zico Kolter" ]
null
null
We introduce Newton-ADMM, a method for fast conic optimization. The basic idea is to view the residuals of consecutive iterates generated by the alternating direction method of multipliers (ADMM) as a set of fixed point equations, and then use a nonsmooth Newton method to find a solution; we apply the basic idea to the...
[]
null
8
1705.00772
title_snapshot
[ -0.07394219189882278, -0.03434136137366295, 0.011145368218421936, 0.007839739322662354, 0.010290304198861122, 0.06292049586772919, 0.00787412654608488, 0.0010436131851747632, -0.04663701727986336, -0.047672126442193985, -0.029259106144309044, -0.0028148312121629715, -0.06463515013456345, -...
Learning Continuous Semantic Representations of Symbolic Expressions
https://proceedings.mlr.press/v70/allamanis17a.html
[ "Miltiadis Allamanis", "Pankajan Chanthirasegaran", "Pushmeet Kohli", "Charles Sutton" ]
null
null
Combining abstract, symbolic reasoning with continuous neural reasoning is a grand challenge of representation learning. As a step in this direction, we propose a new architecture, called neural equivalence network, for the problem of learning continuous semantic representations of algebraic and logical expressions. Th...
[]
null
9
1611.01423
title_snapshot
[ -0.03286636620759964, 0.004355523269623518, -0.01812651753425598, 0.024105042219161987, 0.0341484509408474, 0.03433629125356674, 0.020800756290555, 0.004243520088493824, -0.02439056895673275, 0.007120949216187, -0.02045327052474022, 0.030944187194108963, -0.059571877121925354, 0.0098577672...
Natasha: Faster Non-Convex Stochastic Optimization via Strongly Non-Convex Parameter
https://proceedings.mlr.press/v70/allen-zhu17a.html
[ "Zeyuan Allen-Zhu" ]
null
null
Given a non-convex function $f(x)$ that is an average of $n$ smooth functions, we design stochastic first-order methods to find its approximate stationary points. The performance of our new methods depend on the smallest (negative) eigenvalue $-\sigma$ of the Hessian. This parameter $\sigma$ captures how strongly non-c...
[]
null
10
1702.00763
title_snapshot
[ -0.05638335272669792, -0.022660711780190468, 0.019389009103178978, 0.013775876723229885, 0.025818461552262306, 0.054991915822029114, 0.02341138757765293, 0.009683886542916298, -0.03510849550366402, -0.04226043447852135, -0.0006537392036989331, -0.013890176080167294, -0.043076932430267334, ...
Doubly Accelerated Methods for Faster CCA and Generalized Eigendecomposition
https://proceedings.mlr.press/v70/allen-zhu17b.html
[ "Zeyuan Allen-Zhu", "Yuanzhi Li" ]
null
null
We study k-GenEV, the problem of finding the top k generalized eigenvectors, and k-CCA, the problem of finding the top k vectors in canonical-correlation analysis. We propose algorithms LazyEV and LazyCCA to solve the two problems with running times linearly dependent on the input size and on k. Furthermore, our algori...
[]
null
11
1607.06017
title_snapshot
[ -0.011026008985936642, -0.006593995727598667, 0.01561673916876316, 0.02504093572497368, 0.021505415439605713, 0.03840998560190201, 0.035339657217264175, 0.018847113475203514, 0.023543287068605423, -0.045522794127464294, 0.008626979775726795, -0.006927604787051678, -0.08758991211652756, 0.0...
Faster Principal Component Regression and Stable Matrix Chebyshev Approximation
https://proceedings.mlr.press/v70/allen-zhu17c.html
[ "Zeyuan Allen-Zhu", "Yuanzhi Li" ]
null
null
We solve principal component regression (PCR), up to a multiplicative accuracy $1+\gamma$, by reducing the problem to $\tilde{O}(\gamma^{-1})$ black-box calls of ridge regression. Therefore, our algorithm does not require any explicit construction of the top principal components, and is suitable for large-scale PCR ins...
[]
null
12
1608.04773
title_snapshot
[ -0.03517072647809982, -0.02218344807624817, 0.0028712190687656403, 0.010471955873072147, 0.022034959867596626, 0.04873637109994888, 0.03350764513015747, -0.02512023039162159, -0.009175577200949192, -0.048777222633361816, -0.004951579496264458, 0.014095933176577091, -0.0807962566614151, 0.0...
Follow the Compressed Leader: Faster Online Learning of Eigenvectors and Faster MMWU
https://proceedings.mlr.press/v70/allen-zhu17d.html
[ "Zeyuan Allen-Zhu", "Yuanzhi Li" ]
null
null
The online problem of computing the top eigenvector is fundamental to machine learning. The famous matrix-multiplicative-weight-update (MMWU) framework solves this online problem and gives optimal regret. However, since MMWU runs very slow due to the computation of matrix exponentials, researchers proposed the follow-t...
[]
null
13
1701.01722
title_snapshot
[ -0.004495593719184399, -0.036853041499853134, 0.016318975016474724, 0.026675615459680557, 0.026309335604310036, 0.009835700504481792, 0.041432030498981476, -0.011903252452611923, -0.024983348324894905, -0.07137317210435867, -0.007016325835138559, -0.010786793194711208, -0.0847499743103981, ...
Near-Optimal Design of Experiments via Regret Minimization
https://proceedings.mlr.press/v70/allen-zhu17e.html
[ "Zeyuan Allen-Zhu", "Yuanzhi Li", "Aarti Singh", "Yining Wang" ]
null
null
We consider computationally tractable methods for the experimental design problem, where k out of n design points of dimension p are selected so that certain optimality criteria are approximately satisfied. Our algorithm finds a $(1+\epsilon)$-approximate optimal design when k is a linear function of p; in contrast, ex...
[]
null
14
null
null
[ -0.028445342555642128, -0.0027910638600587845, -0.008150302805006504, 0.05290377885103226, 0.05930451303720474, 0.039022255688905716, 0.012956010177731514, -0.021242395043373108, 0.005208204500377178, -0.06114044040441513, -0.015605962835252285, -0.02198833040893078, -0.03714514151215553, ...
OptNet: Differentiable Optimization as a Layer in Neural Networks
https://proceedings.mlr.press/v70/amos17a.html
[ "Brandon Amos", "J. Zico Kolter" ]
null
null
This paper presents OptNet, a network architecture that integrates optimization problems (here, specifically in the form of quadratic programs) as individual layers in larger end-to-end trainable deep networks. These layers encode constraints and complex dependencies between the hidden states that traditional convoluti...
[]
null
15
1703.00443
title_snapshot
[ -0.04394908621907234, -0.015269470401108265, 0.010943751782178879, 0.031236859038472176, 0.024969158694148064, 0.05643228068947792, -0.012175409123301506, -0.012157815508544445, -0.024512866511940956, -0.02110949158668518, 0.005994631443172693, -0.0011634862748906016, -0.04602326452732086, ...
Input Convex Neural Networks
https://proceedings.mlr.press/v70/amos17b.html
[ "Brandon Amos", "Lei Xu", "J. Zico Kolter" ]
null
null
This paper presents the input convex neural network architecture. These are scalar-valued (potentially deep) neural networks with constraints on the network parameters such that the output of the network is a convex function of (some of) the inputs. The networks allow for efficient inference via optimization over some ...
[]
null
16
1609.07152
title_snapshot
[ -0.025514105334877968, -0.020249582827091217, -0.023781999945640564, 0.059667639434337616, 0.0251931045204401, 0.0622481144964695, -0.012695301324129105, 0.016440391540527344, -0.04457869008183479, -0.035277336835861206, -0.01565355248749256, 0.003499634563922882, -0.049334704875946045, -0...
An Efficient, Sparsity-Preserving, Online Algorithm for Low-Rank Approximation
https://proceedings.mlr.press/v70/anderson17a.html
[ "David Anderson", "Ming Gu" ]
null
null
Low-rank matrix approximation is a fundamental tool in data analysis for processing large datasets, reducing noise, and finding important signals. In this work, we present a novel truncated LU factorization called Spectrum-Revealing LU (SRLU) for effective low-rank matrix approximation, and develop a fast algorithm to ...
[]
null
17
1602.05950
title_snapshot
[ -0.014904890209436417, -0.010556384921073914, 0.015301521867513657, 0.001773558440618217, 0.04157687723636627, 0.0293312668800354, 0.012331916950643063, -0.04710165783762932, -0.04528627172112465, -0.04159916937351227, 0.006372676230967045, 0.005859782453626394, -0.06316840648651123, 0.001...
Modular Multitask Reinforcement Learning with Policy Sketches
https://proceedings.mlr.press/v70/andreas17a.html
[ "Jacob Andreas", "Dan Klein", "Sergey Levine" ]
null
null
We describe a framework for multitask deep reinforcement learning guided by policy sketches. Sketches annotate tasks with sequences of named subtasks, providing information about high-level structural relationships among tasks but not how to implement them—specifically not providing the detailed guidance used by much p...
[]
null
18
1611.01796
title_snapshot
[ -0.0045050461776554585, -0.03077441081404686, -0.002917215693742037, 0.06077498197555542, 0.045250337570905685, 0.0312768816947937, 0.010180084966123104, -0.020361481234431267, -0.040460165590047836, -0.05716912820935249, -0.013947458937764168, 0.020123019814491272, -0.08057364076375961, -...
Averaged-DQN: Variance Reduction and Stabilization for Deep Reinforcement Learning
https://proceedings.mlr.press/v70/anschel17a.html
[ "Oron Anschel", "Nir Baram", "Nahum Shimkin" ]
null
null
Instability and variability of Deep Reinforcement Learning (DRL) algorithms tend to adversely affect their performance. Averaged-DQN is a simple extension to the DQN algorithm, based on averaging previously learned Q-values estimates, which leads to a more stable training procedure and improved performance by reducing ...
[]
null
19
1611.01929
title_snapshot
[ -0.017969949170947075, -0.013235132209956646, -0.03793768957257271, 0.051847148686647415, 0.03633313253521919, 0.016949746757745743, 0.027071645483374596, -0.027379395440220833, -0.05264114588499069, -0.04186400771141052, -0.014516713097691536, -0.005130668170750141, -0.06398297846317291, ...
A Simple Multi-Class Boosting Framework with Theoretical Guarantees and Empirical Proficiency
https://proceedings.mlr.press/v70/appel17a.html
[ "Ron Appel", "Pietro Perona" ]
null
null
There is a need for simple yet accurate white-box learning systems that train quickly and with little data. To this end, we showcase REBEL, a multi-class boosting method, and present a novel family of weak learners called localized similarities. Our framework provably minimizes the training error of any dataset at an e...
[]
null
20
null
null
[ -0.009455373510718346, -0.054554082453250885, -0.00424921466037631, 0.07931866496801376, 0.026146920397877693, 0.03966812044382095, 0.01959133893251419, -0.012022335082292557, 0.00887058675289154, -0.02844955585896969, -0.003961245063692331, 0.035271789878606796, -0.0862116664648056, -0.03...
Deep Voice: Real-time Neural Text-to-Speech
https://proceedings.mlr.press/v70/arik17a.html
[ "Sercan Ö. Arık", "Mike Chrzanowski", "Adam Coates", "Gregory Diamos", "Andrew Gibiansky", "Yongguo Kang", "Xian Li", "John Miller", "Andrew Ng", "Jonathan Raiman", "Shubho Sengupta", "Mohammad Shoeybi" ]
null
null
We present Deep Voice, a production-quality text-to-speech system constructed entirely from deep neural networks. Deep Voice lays the groundwork for truly end-to-end neural speech synthesis. The system comprises five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conv...
[]
null
21
1702.07825
title_snapshot
[ -0.027791477739810944, -0.029528651386499405, -0.01809069514274597, 0.04734495282173157, 0.019145812839269638, 0.05740826949477196, 0.03571246191859245, 0.03360925614833832, -0.017045222222805023, -0.04710932821035385, -0.015767445787787437, 0.019269419834017754, -0.04359418526291847, 0.00...
Oracle Complexity of Second-Order Methods for Finite-Sum Problems
https://proceedings.mlr.press/v70/arjevani17a.html
[ "Yossi Arjevani", "Ohad Shamir" ]
null
null
Finite-sum optimization problems are ubiquitous in machine learning, and are commonly solved using first-order methods which rely on gradient computations. Recently, there has been growing interest insecond-ordermethods, which rely on both gradients and Hessians. In principle, second-order methods can require much fewe...
[]
null
22
1611.04982
title_snapshot
[ -0.06013099104166031, -0.01108074001967907, 0.004372448660433292, 0.01577514223754406, 0.029128167778253555, 0.04429168254137039, 0.020408248528838158, -0.014942293986678123, -0.010445419698953629, -0.04221588000655174, 0.006304699461907148, 0.025672605261206627, -0.060299940407276154, -0....
Wasserstein Generative Adversarial Networks
https://proceedings.mlr.press/v70/arjovsky17a.html
[ "Martin Arjovsky", "Soumith Chintala", "Léon Bottou" ]
null
null
We introduce a new algorithm named WGAN, an alternative to traditional GAN training. In this new model, we show that we can improve the stability of learning, get rid of problems like mode collapse, and provide meaningful learning curves useful for debugging and hyperparameter searches. Furthermore, we show that the co...
[]
null
23
null
null
[ -0.00853447150439024, -0.021496498957276344, 0.0077523100189864635, 0.037485312670469284, 0.007812384981662035, 0.008605990558862686, 0.014196435920894146, 0.013485919684171677, 0.012455676682293415, -0.06158091500401497, -0.015886614099144936, 0.0010323094902560115, -0.05594537779688835, ...
Generalization and Equilibrium in Generative Adversarial Nets (GANs)
https://proceedings.mlr.press/v70/arora17a.html
[ "Sanjeev Arora", "Rong Ge", "Yingyu Liang", "Tengyu Ma", "Yi Zhang" ]
null
null
It is shown that training of generative adversarial network (GAN) may not have good generalization properties; e.g., training may appear successful but the trained distribution may be far from target distribution in standard metrics. However, generalization does occur for a weaker metric called neural net distance. It ...
[]
null
24
1703.00573
title_snapshot
[ -0.019298572093248367, -0.029169172048568726, -0.009564047679305077, 0.03371544927358627, 0.001348564401268959, 0.0144799267873168, 0.007297834847122431, 0.012406590394675732, -0.002805765252560377, -0.05442814528942108, -0.015424733981490135, 0.016070213168859482, -0.09440284967422485, -0...
A Closer Look at Memorization in Deep Networks
https://proceedings.mlr.press/v70/arpit17a.html
[ "Devansh Arpit", "Stanisław Jastrzębski", "Nicolas Ballas", "David Krueger", "Emmanuel Bengio", "Maxinder S. Kanwal", "Tegan Maharaj", "Asja Fischer", "Aaron Courville", "Yoshua Bengio", "Simon Lacoste-Julien" ]
null
null
We examine the role of memorization in deep learning, drawing connections to capacity, generalization, and adversarial robustness. While deep networks are capable of memorizing noise data, our results suggest that they tend to prioritize learning simple patterns first. In our experiments, we expose qualitative differen...
[]
null
25
1706.05394
title_snapshot
[ -0.018390081822872162, 0.0035622301511466503, -0.007230211049318314, 0.06454750895500183, 0.04796503856778145, -0.00414850190281868, 0.043774232268333435, 0.0014105235459282994, -0.037345051765441895, -0.02989552542567253, 0.00031552379368804395, 0.020865632221102715, -0.05821307376027107, ...
An Alternative Softmax Operator for Reinforcement Learning
https://proceedings.mlr.press/v70/asadi17a.html
[ "Kavosh Asadi", "Michael L. Littman" ]
null
null
A softmax operator applied to a set of values acts somewhat like the maximization function and somewhat like an average. In sequential decision making, softmax is often used in settings where it is necessary to maximize utility but also to hedge against problems that arise from putting all of one’s weight behind a sing...
[]
null
26
1612.05628
title_snapshot
[ -0.05818384885787964, -0.022215817123651505, 0.007639470975846052, 0.007700042333453894, 0.051823534071445465, -0.0043831393122673035, 0.011818313971161842, 0.012543349526822567, -0.020177116617560387, -0.06614062190055847, -0.018919052556157112, 0.016000928357243538, -0.0688876137137413, ...
Random Fourier Features for Kernel Ridge Regression: Approximation Bounds and Statistical Guarantees
https://proceedings.mlr.press/v70/avron17a.html
[ "Haim Avron", "Michael Kapralov", "Cameron Musco", "Christopher Musco", "Ameya Velingker", "Amir Zandieh" ]
null
null
Random Fourier features is one of the most popular techniques for scaling up kernel methods, such as kernel ridge regression. However, despite impressive empirical results, the statistical properties of random Fourier features are still not well understood. In this paper we take steps toward filling this gap. Specifica...
[]
null
27
1804.09893
title_snapshot
[ -0.04116528481245041, -0.021668296307325363, 0.02237812615931034, 0.020984528586268425, 0.04308785870671272, 0.0493808388710022, 0.026896940544247627, -0.02976764366030693, -0.03500198945403099, -0.03758608549833298, -0.0289692971855402, 0.011939622461795807, -0.04537581652402878, 0.008816...
Minimax Regret Bounds for Reinforcement Learning
https://proceedings.mlr.press/v70/azar17a.html
[ "Mohammad Gheshlaghi Azar", "Ian Osband", "Rémi Munos" ]
null
null
We consider the problem of provably optimal exploration in reinforcement learning for finite horizon MDPs. We show that an optimistic modification to value iteration achieves a regret bound of $\tilde {O}( \sqrt{HSAT} + H^2S^2A+H\sqrt{T})$ where $H$ is the time horizon, $S$ the number of states, $A$ the number of actio...
[]
null
28
1703.05449
title_snapshot
[ -0.06481589376926422, -0.008491684682667255, -0.00883357785642147, 0.05598945915699005, 0.056390658020973206, 0.022278787568211555, 0.02260606363415718, -0.005134734325110912, -0.012355848215520382, -0.046163614839315414, -0.01257170271128416, -0.004525536671280861, -0.06587541103363037, -...
Learning the Structure of Generative Models without Labeled Data
https://proceedings.mlr.press/v70/bach17a.html
[ "Stephen H. Bach", "Bryan He", "Alexander Ratner", "Christopher Ré" ]
null
null
Curating labeled training data has become the primary bottleneck in machine learning. Recent frameworks address this bottleneck with generative models to synthesize labels at scale from weak supervision sources. The generative model’s dependency structure directly affects the quality of the estimated labels, but select...
[]
null
29
1703.00854
title_snapshot
[ 0.0048815226182341576, -0.03509388491511345, -0.015332415699958801, 0.04253007471561432, 0.03356052190065384, 0.008988477289676666, 0.010566438548266888, -0.030196979641914368, -0.008032996207475662, -0.0185935590416193, 0.0007457792526111007, 0.03126993402838707, -0.07105002552270889, 0.0...
Uniform Deviation Bounds for k-Means Clustering
https://proceedings.mlr.press/v70/bachem17a.html
[ "Olivier Bachem", "Mario Lucic", "S. Hamed Hassani", "Andreas Krause" ]
null
null
Uniform deviation bounds limit the difference between a model’s expected loss and its loss on an empirical sampleuniformlyfor all models in a learning problem. In this paper, we provide a novel framework to obtain uniform deviation bounds for loss functions which areunbounded. As a result, we obtain competitive uniform...
[]
null
30
1702.08249
title_judge
[ -0.012976204045116901, -0.01384661253541708, -0.005038464907556772, 0.03155631199479103, 0.053546153008937836, 0.014068348333239555, 0.041576679795980453, -0.011350994929671288, -0.001886518206447363, -0.01276191882789135, -0.019657032564282417, -0.035515736788511276, -0.059424467384815216, ...
Distributed and Provably Good Seedings for k-Means in Constant Rounds
https://proceedings.mlr.press/v70/bachem17b.html
[ "Olivier Bachem", "Mario Lucic", "Andreas Krause" ]
null
null
The k-Means++ algorithm is the state of the art algorithm to solve k-Means clustering problems as the computed clusterings are O(log k) competitive in expectation. However, its seeding step requires k inherently sequential passes through the full data set making it hard to scale to massive data sets. The standard remed...
[]
null
31
null
null
[ 0.004327867645770311, -0.022685974836349487, 0.04256828874349594, 0.042852871119976044, 0.06001981347799301, 0.032250333577394485, 0.030693093314766884, -0.005569301545619965, -0.01159487385302782, -0.02812371216714382, -0.005050692707300186, -0.05780438706278801, -0.044723015278577805, -0...
Learning Algorithms for Active Learning
https://proceedings.mlr.press/v70/bachman17a.html
[ "Philip Bachman", "Alessandro Sordoni", "Adam Trischler" ]
null
null
We introduce a model that learns active learning algorithms via metalearning. For a distribution of related tasks, our model jointly learns: a data representation, an item selection heuristic, and a prediction function. Our model uses the item selection heuristic to construct a labeled support set for training the pred...
[]
null
32
1708.00088
title_snapshot
[ -0.000013660418517247308, -0.007503668311983347, -0.01148306392133236, 0.024017496034502983, 0.029780114069581032, -0.0031093826983124018, 0.0013196375221014023, -0.0030715488828718662, -0.01853829063475132, 0.0046830023638904095, -0.013247855938971043, 0.051049161702394485, -0.0597707293927...
Improving Viterbi is Hard: Better Runtimes Imply Faster Clique Algorithms
https://proceedings.mlr.press/v70/backurs17a.html
[ "Arturs Backurs", "Christos Tzamos" ]
null
null
The classic algorithm of Viterbi computes the most likely path in a Hidden Markov Model (HMM) that results in a given sequence of observations. It runs in time $O(Tn^2)$ given a sequence of T observations from a HMM with n states. Despite significant interest in the problem and prolonged effort by different communities...
[]
null
33
1607.04229
title_snapshot
[ -0.003356428351253271, -0.018304672092199326, -0.010335570201277733, 0.04234039783477783, 0.06024141609668732, 0.022083278745412827, 0.053861040621995926, 0.04354263097047806, -0.0077935755252838135, -0.06283564120531082, -0.0019347643246874213, -0.017774662002921104, -0.08918847888708115, ...
Differentially Private Clustering in High-Dimensional Euclidean Spaces
https://proceedings.mlr.press/v70/balcan17a.html
[ "Maria-Florina Balcan", "Travis Dick", "Yingyu Liang", "Wenlong Mou", "Hongyang Zhang" ]
null
null
We study the problem of clustering sensitive data while preserving the privacy of individuals represented in the dataset, which has broad applications in practical machine learning and data analysis tasks. Although the problem has been widely studied in the context of low-dimensional, discrete spaces, much remains unkn...
[]
null
34
null
null
[ 0.002415116410702467, 0.007971160113811493, 0.008654460310935974, 0.05245763063430786, 0.05475381016731262, 0.031922366470098495, 0.04357234388589859, -0.03808380290865898, -0.018736911937594414, -0.035033270716667175, -0.009128861129283905, -0.03069738671183586, -0.04559643939137459, 0.01...
Strongly-Typed Agents are Guaranteed to Interact Safely
https://proceedings.mlr.press/v70/balduzzi17a.html
[ "David Balduzzi" ]
null
null
As artificial agents proliferate, it is becoming increasingly important to ensure that their interactions with one another are well-behaved. In this paper, we formalize a common-sense notion of when algorithms are well-behaved: an algorithm is safe if it does no harm. Motivated by recent progress in deep learning, we f...
[]
null
35
1702.07450
title_snapshot
[ -0.06079213693737984, -0.016422942280769348, 0.013330532237887383, 0.015852170065045357, 0.01666206680238247, 0.004830397665500641, 0.03251991793513298, 0.0046769059263169765, -0.013346326537430286, -0.04301512613892555, -0.03071345016360283, 0.03858677297830582, -0.0997675284743309, -0.00...
The Shattered Gradients Problem: If resnets are the answer, then what is the question?
https://proceedings.mlr.press/v70/balduzzi17b.html
[ "David Balduzzi", "Marcus Frean", "Lennox Leary", "J. P. Lewis", "Kurt Wan-Duo Ma", "Brian McWilliams" ]
null
null
A long-standing obstacle to progress in deep learning is the problem of vanishing and exploding gradients. Although, the problem has largely been overcome via carefully constructed initializations and batch normalization, architectures incorporating skip-connections such as highway and resnets perform much better than ...
[]
null
36
1702.08591
title_snapshot
[ 0.00616056052967906, -0.016960810869932175, -0.006173679605126381, 0.06792397052049637, 0.0441678911447525, 0.044938236474990845, 0.022157279774546623, 0.017554279416799545, -0.009663821198046207, -0.03743814677000046, -0.004357971716672182, 0.002519025234505534, -0.04628044739365578, -0.0...
Neural Taylor Approximations: Convergence and Exploration in Rectifier Networks
https://proceedings.mlr.press/v70/balduzzi17c.html
[ "David Balduzzi", "Brian McWilliams", "Tony Butler-Yeoman" ]
null
null
Modern convolutional networks, incorporating rectifiers and max-pooling, are neither smooth nor convex; standard guarantees therefore do not apply. Nevertheless, methods from convex optimization such as gradient descent and Adam are widely used as building blocks for deep learning algorithms. This paper provides the fi...
[]
null
37
1611.02345
title_snapshot
[ -0.0340178906917572, -0.027876287698745728, 0.007050883956253529, 0.011129685677587986, 0.026311328634619713, 0.04251394793391228, 0.02331089787185192, 0.008285054005682468, -0.03128649666905403, -0.024724140763282776, 0.005670254584401846, 0.021286966279149055, -0.06025488302111626, -0.01...
Spectral Learning from a Single Trajectory under Finite-State Policies
https://proceedings.mlr.press/v70/balle17a.html
[ "Borja Balle", "Odalric-Ambrym Maillard" ]
null
null
We present spectral methods of moments for learning sequential models from a single trajectory, in stark contrast with the classical literature that assumes the availability of multiple i.i.d. trajectories. Our approach leverages an efficient SVD-based learning algorithm for weighted automata and provides the first rig...
[]
null
38
null
null
[ -0.034633003175258636, -0.01904054917395115, -0.01566823199391365, 0.03024626336991787, 0.032380443066358566, 0.028677450492978096, 0.04078083857893944, -0.0007082811207510531, 0.006408076733350754, -0.04111596569418907, 0.024053756147623062, -0.007025898899883032, -0.07730378210544586, -0...
Lost Relatives of the Gumbel Trick
https://proceedings.mlr.press/v70/balog17a.html
[ "Matej Balog", "Nilesh Tripuraneni", "Zoubin Ghahramani", "Adrian Weller" ]
null
null
The Gumbel trick is a method to sample from a discrete probability distribution, or to estimate its normalizing partition function. The method relies on repeatedly applying a random perturbation to the distribution in a particular way, each time solving for the most likely configuration. We derive an entire family of r...
[]
null
39
1706.04161
title_snapshot
[ -0.03147788345813751, -0.026247022673487663, -0.00873608235269785, 0.03935297951102257, 0.0540052205324173, 0.000814863364212215, 0.02739967592060566, -0.004958200268447399, -0.03246511518955231, -0.05977853760123253, -0.0016766447806730866, -0.007720855996012688, -0.07661407440900803, -0....
Dynamic Word Embeddings
https://proceedings.mlr.press/v70/bamler17a.html
[ "Robert Bamler", "Stephan Mandt" ]
null
null
We present a probabilistic language model for time-stamped text data which tracks the semantic evolution of individual words over time. The model represents words and contexts by latent trajectories in an embedding space. At each moment in time, the embedding vectors are inferred from a probabilistic version of word2ve...
[]
null
40
1702.08359
title_snapshot
[ 0.005395687185227871, -0.015201826579868793, -0.0011558575788512826, 0.04908876121044159, 0.044159479439258575, 0.03080291487276554, 0.03757807984948158, 0.03665859252214432, 0.00534701906144619, -0.011047447100281715, 0.003937835339456797, 0.00391196645796299, -0.027710292488336563, 0.009...
End-to-End Differentiable Adversarial Imitation Learning
https://proceedings.mlr.press/v70/baram17a.html
[ "Nir Baram", "Oron Anschel", "Itai Caspi", "Shie Mannor" ]
null
null
Generative Adversarial Networks (GANs) have been successfully applied to the problem ofpolicy imitationin a model-free setup. However, the computation graph of GANs, that include a stochastic policy as the generative model, is no longer differentiable end-to-end, which requires the use of high-variance gradient estimat...
[]
null
41
null
null
[ -0.001981648849323392, -0.024813640862703323, -0.013864811509847641, 0.04089997708797455, 0.022371847182512283, 0.016655687242746353, 0.02872750163078308, 0.010122019797563553, -0.006645433604717255, -0.04725733399391174, -0.01671036146581173, -0.01287104468792677, -0.06951909512281418, 0....
Emulating the Expert: Inverse Optimization through Online Learning
https://proceedings.mlr.press/v70/barmann17a.html
[ "Andreas Bärmann", "Sebastian Pokutta", "Oskar Schneider" ]
null
null
In this paper, we demonstrate how to learn the objective function of a decision maker while only observing the problem input data and the decision maker’s corresponding decisions over multiple rounds. Our approach is based on online learning techniques and works for linear objectives over arbitrary sets for which we ha...
[]
null
42
null
null
[ -0.028303490951657295, 0.0011698699090629816, 0.0036571738310158253, 0.034071169793605804, 0.057097699493169785, 0.052210476249456406, 0.005240800324827433, 0.008514299057424068, -0.00576811982318759, -0.01666209287941456, -0.02420836314558983, 0.023068789392709732, -0.07179354876279831, -...
Unimodal Probability Distributions for Deep Ordinal Classification
https://proceedings.mlr.press/v70/beckham17a.html
[ "Christopher Beckham", "Christopher Pal" ]
null
null
Probability distributions produced by the cross-entropy loss for ordinal classification problems can possess undesired properties. We propose a straightforward technique to constrain discrete ordinal probability distributions to be unimodal via the use of the Poisson and binomial probability distributions. We evaluate ...
[]
null
43
1705.05278
title_snapshot
[ -0.018817605450749397, -0.0315290242433548, -0.002283997368067503, 0.016180897131562233, 0.03321056813001633, 0.04259706661105156, 0.0042805979028344154, 0.001065893447957933, -0.02449401468038559, -0.038550276309251785, -0.004737245384603739, 0.005128653720021248, -0.06297335028648376, -0...
Globally Induced Forest: A Prepruning Compression Scheme
https://proceedings.mlr.press/v70/begon17a.html
[ "Jean-Michel Begon", "Arnaud Joly", "Pierre Geurts" ]
null
null
Tree-based ensemble models are heavy memory-wise. An undesired state of affairs considering nowadays datasets, memory-constrained environment and fitting/prediction times. In this paper, we propose the Globally Induced Forest (GIF) to remedy this problem. GIF is a fast prepruning approach to build lightweight ensembles...
[]
null
44
null
null
[ -0.0197260994464159, -0.055184632539749146, -0.006186723243445158, 0.02396250329911709, 0.03872241452336311, 0.027910351753234863, 0.031206516548991203, 0.0005138654378242791, -0.04019290208816528, -0.03154069185256958, -0.004885210655629635, -0.025061247870326042, -0.09874601662158966, -0...
End-to-End Learning for Structured Prediction Energy Networks
https://proceedings.mlr.press/v70/belanger17a.html
[ "David Belanger", "Bishan Yang", "Andrew McCallum" ]
null
null
Structured Prediction Energy Networks (SPENs) are a simple, yet expressive family of structured prediction models (Belanger and McCallum, 2016). An energy function over candidate structured outputs is given by a deep network, and predictions are formed by gradient-based optimization. This paper presents end-to-end lear...
[]
null
45
1703.05667
title_snapshot
[ 0.008018994703888893, -0.021011853590607643, 0.012369534932076931, 0.030515771359205246, 0.050630275160074234, 0.009919906035065651, 0.00009304474224336445, -0.013674665242433548, -0.022834412753582, -0.04286804795265198, 0.005362243391573429, 0.015365004539489746, -0.052916813641786575, 0...
Learning to Discover Sparse Graphical Models
https://proceedings.mlr.press/v70/belilovsky17a.html
[ "Eugene Belilovsky", "Kyle Kastner", "Gael Varoquaux", "Matthew B. Blaschko" ]
null
null
We consider structure discovery of undirected graphical models from observational data. Inferring likely structures from few examples is a complex task often requiring the formulation of priors and sophisticated inference procedures. Popular methods rely on estimating a penalized maximum likelihood of the precision mat...
[]
null
46
1605.06359
title_snapshot
[ -0.012722140178084373, -0.004215385764837265, -0.0007409955142065883, 0.039149969816207886, 0.03546485677361488, 0.02960110828280449, 0.04080805554986, 0.01627875305712223, -0.015526525676250458, -0.05150563642382622, 0.040557075291872025, 0.02240031585097313, -0.05987183377146721, -0.0039...
A Distributional Perspective on Reinforcement Learning
https://proceedings.mlr.press/v70/bellemare17a.html
[ "Marc G. Bellemare", "Will Dabney", "Rémi Munos" ]
null
null
In this paper we argue for the fundamental importance of the value distribution: the distribution of the random return received by a reinforcement learning agent. This is in contrast to the common approach to reinforcement learning which models the expectation of this return, or value. Although there is an established ...
[]
null
47
1707.06887
title_snapshot
[ -0.015035219490528107, -0.015817519277334213, -0.007732538040727377, 0.032198768109083176, 0.05313645675778389, 0.030139632523059845, -0.007729826495051384, 0.0047364928759634495, -0.02663484588265419, -0.04751363769173622, -0.012270329520106316, 0.007549038622528315, -0.06924266368150711, ...
Neural Optimizer Search with Reinforcement Learning
https://proceedings.mlr.press/v70/bello17a.html
[ "Irwan Bello", "Barret Zoph", "Vijay Vasudevan", "Quoc V. Le" ]
null
null
We present an approach to automate the process of discovering optimization methods, with a focus on deep learning architectures. We train a Recurrent Neural Network controller to generate a string in a specific domain language that describes a mathematical update equation based on a list of primitive functions, such as...
[]
null
48
1709.07417
title_snapshot
[ -0.035348694771528244, -0.03041732870042324, -0.005657818168401718, 0.056052759289741516, 0.049559131264686584, 0.06379521638154984, 0.0009198729530908167, 0.009309564717113972, -0.010990045964717865, -0.046021297574043274, -0.027005283161997795, 0.023104287683963776, -0.048324014991521835, ...
Learning Texture Manifolds with the Periodic Spatial GAN
https://proceedings.mlr.press/v70/bergmann17a.html
[ "Urs Bergmann", "Nikolay Jetchev", "Roland Vollgraf" ]
null
null
This paper introduces a novel approach to texture synthesis based on generative adversarial networks (GAN) (Goodfellow et al., 2014), and call this technique Periodic Spatial GAN (PSGAN). The PSGAN has several novel abilities which surpass the current state of the art in texture synthesis. First, we can learn multiple ...
[]
null
49
1705.06566
title_snapshot
[ 0.00032344370265491307, -0.006036947015672922, -0.007125332485884428, 0.020362764596939087, 0.008809475228190422, 0.017248056828975677, 0.007943985052406788, 0.019891902804374695, -0.019920453429222107, -0.0739411935210228, -0.03424447774887085, -0.028899194672703743, -0.054702285677194595, ...
Differentially Private Learning of Undirected Graphical Models Using Collective Graphical Models
https://proceedings.mlr.press/v70/bernstein17a.html
[ "Garrett Bernstein", "Ryan McKenna", "Tao Sun", "Daniel Sheldon", "Michael Hay", "Gerome Miklau" ]
null
null
We investigate the problem of learning discrete graphical models in a differentially private way. Approaches to this problem range from privileged algorithms that conduct learning completely behind the privacy barrier to schemes that release private summary statistics paired with algorithms to learn parameters from tho...
[]
null
50
1706.04646
title_snapshot
[ -0.0009953109547495842, 0.011396098881959915, -0.014446474611759186, 0.04252171516418457, 0.052953243255615234, 0.00007541523518739268, 0.05372704938054085, -0.020291993394494057, 0.0029060342349112034, -0.036790139973163605, 0.01599244400858879, 0.003175661200657487, -0.07259529829025269, ...
Efficient Online Bandit Multiclass Learning with $\tilde{O}(\sqrt{T})$ Regret
https://proceedings.mlr.press/v70/beygelzimer17a.html
[ "Alina Beygelzimer", "Francesco Orabona", "Chicheng Zhang" ]
null
null
We present an efficient second-order algorithm with $\tilde{O}(1/\eta \sqrt{T})$ regret for the bandit online multiclass problem. The regret bound holds simultaneously with respect to a family of loss functions parameterized by $\eta$, ranging from hinge loss ($\eta=0$) to squared hinge loss ($\eta=1$). This provides a...
[]
null
51
1702.07958
title_snapshot
[ -0.020225889980793, -0.004748079460114241, 0.007030921056866646, 0.04762618616223335, 0.021058982238173485, 0.040367186069488525, 0.020710177719593048, 0.0020892727188766003, -0.01671317033469677, -0.05038866400718689, -0.008199851959943771, 0.006889073643833399, -0.06708282977342606, -0.0...
Guarantees for Greedy Maximization of Non-submodular Functions with Applications
https://proceedings.mlr.press/v70/bian17a.html
[ "Andrew An Bian", "Joachim M. Buhmann", "Andreas Krause", "Sebastian Tschiatschek" ]
null
null
We investigate the performance of the standard Greedy algorithm for cardinality constrained maximization of non-submodular nondecreasing set functions. While there are strong theoretical guarantees on the performance of Greedy for maximizing submodular functions, there are few guarantees for non-submodular ones. Howeve...
[]
null
52
1703.02100
title_snapshot
[ -0.029973655939102173, -0.010928655043244362, 0.011320352554321289, 0.054097067564725876, 0.04516753926873207, 0.04379269853234291, 0.01066158153116703, -0.0397518090903759, -0.03635858744382858, -0.040382687002420425, -0.0260329507291317, 0.021379228681325912, -0.06991660594940186, -0.020...
Robust Submodular Maximization: A Non-Uniform Partitioning Approach
https://proceedings.mlr.press/v70/bogunovic17a.html
[ "Ilija Bogunovic", "Slobodan Mitrović", "Jonathan Scarlett", "Volkan Cevher" ]
null
null
We study the problem of maximizing a monotone submodular function subject to a cardinality constraint $k$, with the added twist that a number of items $\tau$ from the returned set may be removed. We focus on the worst-case setting considered by Orlin et al.\ (2016), in which a constant-factor approximation guarantee wa...
[]
null
53
1706.04918
title_snapshot
[ -0.0018964960472658277, -0.014870806597173214, -0.005864471197128296, 0.05031546205282211, 0.060545288026332855, 0.05649144574999809, 0.005389539059251547, -0.029592731967568398, -0.01835421286523342, -0.04641937464475632, -0.0017041023820638657, -0.020482264459133148, -0.07589410245418549, ...
Unsupervised Learning by Predicting Noise
https://proceedings.mlr.press/v70/bojanowski17a.html
[ "Piotr Bojanowski", "Armand Joulin" ]
null
null
Convolutional neural networks provide visual features that perform remarkably well in many computer vision applications. However, training these networks requires significant amounts of supervision; this paper introduces a generic framework to train such networks, end-to-end, with no supervision. We propose to fix a se...
[]
null
54
1704.05310
title_snapshot
[ 0.029188113287091255, -0.02351987548172474, -0.027502357959747314, 0.022109482437372208, 0.009330940432846546, 0.03969282656908035, 0.013929158449172974, 0.01036739069968462, -0.012235959991812706, -0.04493890702724457, -0.023547474294900894, 0.006298070307821035, -0.07227694988250732, 0.0...
Adaptive Neural Networks for Efficient Inference
https://proceedings.mlr.press/v70/bolukbasi17a.html
[ "Tolga Bolukbasi", "Joseph Wang", "Ofer Dekel", "Venkatesh Saligrama" ]
null
null
We present an approach to adaptively utilize deep neural networks in order to reduce the evaluation time on new examples without loss of accuracy. Rather than attempting to redesign or approximate existing networks, we propose two schemes that adaptively utilize networks. We first pose an adaptive network evaluation sc...
[]
null
55
1702.07811
title_snapshot
[ -0.021311640739440918, -0.023979557678103447, 0.00615659449249506, 0.05179499089717865, 0.04192906990647316, 0.05017101392149925, 0.0038304973859339952, 0.0036996938288211823, -0.02248351089656353, -0.045573458075523376, -0.004435394424945116, 0.0027619251050055027, -0.04240429028868675, -...
Compressed Sensing using Generative Models
https://proceedings.mlr.press/v70/bora17a.html
[ "Ashish Bora", "Ajil Jalal", "Eric Price", "Alexandros G. Dimakis" ]
null
null
The goal of compressed sensing is to estimate a vector from an underdetermined system of noisy linear measurements, by making use of prior knowledge on the structure of vectors in the relevant domain. For almost all results in this literature, the structure is represented by sparsity in a well-chosen basis. We show how...
[]
null
56
1703.03208
title_snapshot
[ 0.006244240328669548, -0.021071359515190125, 0.003015862312167883, 0.04462243244051933, 0.04159940779209137, 0.035630084574222565, 0.04303461313247681, 0.0022281298879534006, -0.03383975476026535, -0.07026517391204834, -0.0027505308389663696, -0.02059054933488369, -0.05138738080859184, 0.0...
Programming with a Differentiable Forth Interpreter
https://proceedings.mlr.press/v70/bosnjak17a.html
[ "Matko Bošnjak", "Tim Rocktäschel", "Jason Naradowsky", "Sebastian Riedel" ]
null
null
Given that in practice training data is scarce for all but a small set of problems, a core question is how to incorporate prior knowledge into a model. In this paper, we consider the case of prior procedural knowledge for neural networks, such as knowing how a program should traverse a sequence, but not what local acti...
[]
null
57
1605.06640
title_snapshot
[ -0.03721994161605835, -0.012865985743701458, -0.034363824874162674, 0.02078137919306755, 0.04287315160036087, 0.03918038681149483, 0.037270672619342804, -0.002156296046450734, -0.0374135822057724, -0.022629201412200928, -0.003472436685115099, 0.030042188242077827, -0.07009661942720413, -0....
Practical Gauss-Newton Optimisation for Deep Learning
https://proceedings.mlr.press/v70/botev17a.html
[ "Aleksandar Botev", "Hippolyt Ritter", "David Barber" ]
null
null
We present an efficient block-diagonal approximation to the Gauss-Newton matrix for feedforward neural networks. Our resulting algorithm is competitive against state-of-the-art first-order optimisation methods, with sometimes significant improvement in optimisation performance. Unlike first-order methods, for which hyp...
[]
null
58
1706.03662
title_snapshot
[ -0.056706853210926056, -0.017565593123435974, 0.027204211801290512, 0.012104644440114498, 0.007706949021667242, 0.0749557614326477, 0.013642816804349422, 0.003934667445719242, -0.028376538306474686, -0.01631312258541584, -0.024152174592018127, 0.010597427375614643, -0.040148261934518814, -...
Lazifying Conditional Gradient Algorithms
https://proceedings.mlr.press/v70/braun17a.html
[ "Gábor Braun", "Sebastian Pokutta", "Daniel Zink" ]
null
null
Conditional gradient algorithms (also often called Frank-Wolfe algorithms) are popular due to their simplicity of only requiring a linear optimization oracle and more recently they also gained significant traction for online learning. While simple in principle, in many cases the actual implementation of the linear opti...
[]
null
59
1610.05120
title_snapshot
[ -0.01011700090020895, 0.0009259490179829299, 0.005024319048970938, 0.017304880544543266, 0.04752171039581299, 0.047610875219106674, 0.01293447520583868, -0.011641070246696472, -0.010389992967247963, -0.01802767626941204, -0.02154514193534851, 0.02150414139032364, -0.057660456746816635, -0....
Clustering High Dimensional Dynamic Data Streams
https://proceedings.mlr.press/v70/braverman17a.html
[ "Vladimir Braverman", "Gereon Frahling", "Harry Lang", "Christian Sohler", "Lin F. Yang" ]
null
null
We present data streaming algorithms for the $k$-median problem in high-dimensional dynamic geometric data streams, i.e. streams allowing both insertions and deletions of points from a discrete Euclidean space $\{1, 2, \ldots \Delta\}^d$. Our algorithms use $k \epsilon^{-2} \mathrm{poly}(d \log \Delta)$ space/time and ...
[]
null
60
1706.03887
title_snapshot
[ -0.019966263324022293, -0.030253941193223, 0.01916836015880108, 0.062305573374032974, 0.03341638669371605, 0.051340702921152115, 0.0040051573887467384, 0.009502124972641468, -0.03272213041782379, -0.0687771886587143, -0.023524947464466095, -0.05558031424880028, -0.06728055328130722, 0.0018...
On the Sampling Problem for Kernel Quadrature
https://proceedings.mlr.press/v70/briol17a.html
[ "François-Xavier Briol", "Chris J. Oates", "Jon Cockayne", "Wilson Ye Chen", "Mark Girolami" ]
null
null
The standard Kernel Quadrature method for numerical integration with random point sets (also called Bayesian Monte Carlo) is known to converge in root mean square error at a rate determined by the ratio s/d, where s and d encode the smoothness and dimension of the integrand. However, an empirical investigation reveals ...
[]
null
61
1706.03369
title_snapshot
[ -0.039692219346761703, -0.002528765005990863, 0.017261909320950508, 0.0680304691195488, 0.03614851087331772, 0.04894624650478363, -0.006809058133512735, -0.026896648108959198, -0.043380655348300934, -0.06166306138038635, -0.01132788136601448, 0.0013237058883532882, -0.0352683961391449, 0.0...
Reduced Space and Faster Convergence in Imperfect-Information Games via Pruning
https://proceedings.mlr.press/v70/brown17a.html
[ "Noam Brown", "Tuomas Sandholm" ]
null
null
Iterative algorithms such as Counterfactual Regret Minimization (CFR) are the most popular way to solve large zero-sum imperfect-information games. In this paper we introduce Best-Response Pruning (BRP), an improvement to iterative algorithms such as CFR that allows poorly-performing actions to be temporarily pruned. W...
[]
null
62
1609.03234
title_judge
[ -0.05549238622188568, -0.024485766887664795, -0.0018009331542998552, 0.03529324010014534, 0.030799083411693573, 0.03638396039605141, -0.003570521716028452, 0.007703596726059914, -0.038999393582344055, -0.05017849802970886, -0.016889577731490135, 0.02723427303135395, -0.07198183983564377, -...
Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs
https://proceedings.mlr.press/v70/brutzkus17a.html
[ "Alon Brutzkus", "Amir Globerson" ]
null
null
Deep learning models are often successfully trained using gradient descent, despite the worst case hardness of the underlying non-convex optimization problem. The key question is then under what conditions can one prove that optimization will succeed. Here we provide a strong result of this kind. We consider a neural n...
[]
null
63
1702.07966
title_snapshot
[ -0.005623030476272106, -0.023603493347764015, 0.01333596557378769, 0.03981985151767731, 0.01858833245933056, 0.03951616957783699, 0.019132345914840698, 0.016387682408094406, -0.016621919348835945, -0.02656230516731739, -0.02747104875743389, 0.007247439119964838, -0.0690738633275032, -0.000...
Deep Tensor Convolution on Multicores
https://proceedings.mlr.press/v70/budden17a.html
[ "David Budden", "Alexander Matveev", "Shibani Santurkar", "Shraman Ray Chaudhuri", "Nir Shavit" ]
null
null
Deep convolutional neural networks (ConvNets) of 3-dimensional kernels allow joint modeling of spatiotemporal features. These networks have improved performance of video and volumetric image analysis, but have been limited in size due to the low memory ceiling of GPU hardware. Existing CPU implementations overcome this...
[]
null
64
1611.06565
title_snapshot
[ -0.0014008809812366962, -0.040689945220947266, 0.004560962785035372, 0.03586006537079811, 0.028448177501559258, 0.03940131887793541, 0.015033163130283356, 0.03848651051521301, -0.01701403595507145, -0.058463841676712036, 0.006297439336776733, -0.01210224349051714, -0.05125678703188896, 0.0...
Multi-objective Bandits: Optimizing the Generalized Gini Index
https://proceedings.mlr.press/v70/busa-fekete17a.html
[ "Róbert Busa-Fekete", "Balázs Szörényi", "Paul Weng", "Shie Mannor" ]
null
null
We study the multi-armed bandit (MAB) problem where the agent receives a vectorial feedback that encodes many possibly competing objectives to be optimized. The goal of the agent is to find a policy, which can optimize these objectives simultaneously in a fair way. This multi-objective online optimization problem is fo...
[]
null
65
1706.04933
title_snapshot
[ -0.04465311020612717, -0.023688893765211105, 0.033234674483537674, 0.035201992839574814, 0.03230053186416626, 0.03987770527601242, 0.024806395173072815, 0.017604639753699303, -0.01910596713423729, -0.04267214611172676, 0.016354478895664215, 0.022797690704464912, -0.09000430256128311, -0.02...
Priv’IT: Private and Sample Efficient Identity Testing
https://proceedings.mlr.press/v70/cai17a.html
[ "Bryan Cai", "Constantinos Daskalakis", "Gautam Kamath" ]
null
null
We develop differentially private hypothesis testing methods for the small sample regime. Given a sample $\mathcal{D}$ from a categorical distribution $p$ over some domain $\Sigma$, an explicitly described distribution $q$ over $\Sigma$, some privacy parameter $\epsilon$, accuracy parameter $\alpha$, and requirements $...
[]
null
66
1703.10127
title_snapshot
[ -0.007080917712301016, 0.015045328065752983, -0.001997280167415738, 0.06471741199493408, 0.059526968747377396, 0.016556791961193085, 0.05440783500671387, -0.03957730159163475, -0.013804817572236061, -0.041351478546857834, 0.0188607145100832, -0.019284751266241074, -0.052570294588804245, 0....
Second-Order Kernel Online Convex Optimization with Adaptive Sketching
https://proceedings.mlr.press/v70/calandriello17a.html
[ "Daniele Calandriello", "Alessandro Lazaric", "Michal Valko" ]
null
null
Kernel online convex optimization (KOCO) is a framework combining the expressiveness of non-parametric kernel models with the regret guarantees of online learning. First-order KOCO methods such as functional gradient descent require only $O(t)$ time and space per iteration, and, when the only information on the losses ...
[]
null
67
1706.04892
title_snapshot
[ -0.05575793981552124, -0.025927480310201645, 0.0428682416677475, 0.044074755162000656, 0.01928727887570858, 0.06458720564842224, -0.0012823648285120726, 0.005247762426733971, 0.0019385224441066384, -0.05204484611749649, -0.015744555741548538, 0.005465724505484104, -0.03725297749042511, -0....
“Convex Until Proven Guilty”: Dimension-Free Acceleration of Gradient Descent on Non-Convex Functions
https://proceedings.mlr.press/v70/carmon17a.html
[ "Yair Carmon", "John C. Duchi", "Oliver Hinder", "Aaron Sidford" ]
null
null
We develop and analyze a variant of Nesterov’s accelerated gradient descent (AGD) for minimization of smooth non-convex functions. We prove that one of two cases occurs: either our AGD variant converges quickly, as if the function was convex, or we produce a certificate that the function is “guilty” of being non-convex...
[]
null
68
1705.02766
title_snapshot
[ -0.051947589963674545, -0.027382150292396545, 0.02090570330619812, 0.03390306979417801, 0.01628798618912697, 0.045683689415454865, 0.022188208997249603, 0.00990865658968687, -0.033116452395915985, -0.0420304499566555, -0.008932605385780334, 0.015449944883584976, -0.049478303641080856, 0.00...
Sliced Wasserstein Kernel for Persistence Diagrams
https://proceedings.mlr.press/v70/carriere17a.html
[ "Mathieu Carrière", "Marco Cuturi", "Steve Oudot" ]
null
null
Persistence diagrams (PDs) play a key role in topological data analysis (TDA), in which they are routinely used to describe succinctly complex topological properties of complicated shapes. PDs enjoy strong stability properties and have proven their utility in various learning contexts. They do not, however, live in a s...
[]
null
69
1706.03358
title_snapshot
[ -0.04553348943591118, -0.02043658308684826, -0.0023862964008003473, 0.057072099298238754, 0.025386106222867966, 0.024083873257040977, 0.01840946264564991, 0.009100069291889668, -0.004542673006653786, -0.04682038724422455, -0.033269695937633514, -0.008481702767312527, -0.06047815456986427, ...
Multiple Clustering Views from Multiple Uncertain Experts
https://proceedings.mlr.press/v70/chang17a.html
[ "Yale Chang", "Junxiang Chen", "Michael H. Cho", "Peter J. Castaldi", "Edwin K. Silverman", "Jennifer G. Dy" ]
null
null
Expert input can improve clustering performance. In today’s collaborative environment, the availability of crowdsourced multiple expert input is becoming common. Given multiple experts’ inputs, most existing approaches can only discover one clustering structure. However, data is multi-faced by nature and can be cluster...
[]
null
70
null
null
[ 0.018602997064590454, -0.0156966932117939, -0.013632375746965408, 0.03667834773659706, 0.03551408648490906, 0.023831883445382118, 0.014005950652062893, -0.026662280783057213, -0.025202538818120956, -0.051198720932006836, -0.021140534430742264, 0.0027581716421991587, -0.06725984066724777, 0...
Uncertainty Assessment and False Discovery Rate Control in High-Dimensional Granger Causal Inference
https://proceedings.mlr.press/v70/chaudhry17a.html
[ "Aditya Chaudhry", "Pan Xu", "Quanquan Gu" ]
null
null
Causal inference among high-dimensional time series data proves an important research problem in many fields. While in the classical regime one often establishes causality among time series via a concept known as “Granger causality,” existing approaches for Granger causal inference in high-dimensional data lack the mea...
[]
null
71
null
null
[ 0.007698928005993366, -0.012276038527488708, -0.020609138533473015, 0.015491296537220478, 0.04336932674050331, 0.0275596734136343, 0.06516177207231522, -0.006148665212094784, -0.019869698211550713, -0.05734748765826225, 0.0027670678682625294, -0.005210024770349264, -0.05509268119931221, 0....
Active Heteroscedastic Regression
https://proceedings.mlr.press/v70/chaudhuri17a.html
[ "Kamalika Chaudhuri", "Prateek Jain", "Nagarajan Natarajan" ]
null
null
An active learner is given a model class $\Theta$, a large sample of unlabeled data drawn from an underlying distribution and access to a labeling oracle that can provide a label for any of the unlabeled instances. The goal of the learner is to find a model $\theta \in \Theta$ that fits the data to a given accuracy whi...
[]
null
72
null
null
[ -0.0012559982715174556, -0.012945079244673252, -0.021760955452919006, 0.008810065686702728, 0.033265698701143265, 0.02843337319791317, 0.015716219320893288, -0.015184126794338226, -0.02268410660326481, -0.020965702831745148, -0.014280329458415508, 0.030044984072446823, -0.07278774678707123, ...
Combining Model-Based and Model-Free Updates for Trajectory-Centric Reinforcement Learning
https://proceedings.mlr.press/v70/chebotar17a.html
[ "Yevgen Chebotar", "Karol Hausman", "Marvin Zhang", "Gaurav Sukhatme", "Stefan Schaal", "Sergey Levine" ]
null
null
Reinforcement learning algorithms for real-world robotic applications must be able to handle complex, unknown dynamical systems while maintaining data-efficient learning. These requirements are handled well by model-free and model-based RL approaches, respectively. In this work, we aim to combine the advantages of thes...
[]
null
73
1703.03078
title_snapshot
[ -0.021000005304813385, -0.005720531102269888, -0.006318013183772564, 0.01928410306572914, 0.047897227108478546, 0.011912361718714237, 0.0036066602915525436, 0.010595486499369144, -0.02547667920589447, -0.049084749072790146, -0.03024805337190628, 0.014455001801252365, -0.08080282807350159, ...
Robust Structured Estimation with Single-Index Models
https://proceedings.mlr.press/v70/chen17a.html
[ "Sheng Chen", "Arindam Banerjee" ]
null
null
In this paper, we investigate general single-index models (SIMs) in high dimensions. Based on U-statistics, we propose two types of robust estimators for the recovery of model parameters, which can be viewed as generalizations of several existing algorithms for one-bit compressed sensing (1-bit CS). With minimal assump...
[]
null
74
null
null
[ -0.01968846656382084, -0.001444538589566946, 0.029966888949275017, 0.006250529550015926, 0.04323360696434975, 0.052748262882232666, 0.0131990946829319, 0.0075139375403523445, -0.043415240943431854, -0.05956804007291794, 0.025378670543432236, -0.02065013349056244, -0.05959475412964821, -0.0...
Adaptive Multiple-Arm Identification
https://proceedings.mlr.press/v70/chen17b.html
[ "Jiecao Chen", "Xi Chen", "Qin Zhang", "Yuan Zhou" ]
null
null
We study the problem of selecting K arms with the highest expected rewards in a stochastic n-armed bandit game. This problem has a wide range of applications, e.g., A/B testing, crowdsourcing, simulation optimization. Our goal is to develop a PAC algorithm, which, with probability at least $1-\delta$, identifies a set ...
[]
null
75
1706.01026
title_snapshot
[ -0.03680397942662239, 0.000890246476046741, 0.008511736989021301, 0.031864289194345474, 0.03533676266670227, 0.036171842366456985, 0.024943534284830093, 0.0014700081665068865, -0.035526297986507416, -0.053165603429079056, 0.005570598412305117, 0.005992827005684376, -0.04680314660072327, -0...
Dueling Bandits with Weak Regret
https://proceedings.mlr.press/v70/chen17c.html
[ "Bangrui Chen", "Peter I. Frazier" ]
null
null
We consider online content recommendation with implicit feedback through pairwise comparisons, formalized as the so-called dueling bandit problem. We study the dueling bandit problem in the Condorcet winner setting, and consider two notions of regret: the more well-studied strong regret, which is 0 only when both arms ...
[]
null
76
1706.04304
title_snapshot
[ -0.02520724944770336, -0.019822590053081512, -0.009128415025770664, 0.051803216338157654, 0.03159746900200844, 0.0010695509845390916, 0.003686051582917571, 0.024336958304047585, -0.0032289845403283834, -0.03936564922332764, -0.007631936576217413, 0.02354160137474537, -0.045901473611593246, ...
Strong NP-Hardness for Sparse Optimization with Concave Penalty Functions
https://proceedings.mlr.press/v70/chen17d.html
[ "Yichen Chen", "Dongdong Ge", "Mengdi Wang", "Zizhuo Wang", "Yinyu Ye", "Hao Yin" ]
null
null
Consider the regularized sparse minimization problem, which involves empirical sums of loss functions for $n$ data points (each of dimension $d$) and a nonconvex sparsity penalty. We prove that finding an $\mathcal{O}(n^{c_1}d^{c_2})$-optimal solution to the regularized sparse optimization problem is strongly NP-hard f...
[]
null
77
1501.00622
title_snapshot
[ -0.024469967931509018, -0.025056857615709305, 0.021308645606040955, 0.026210594922304153, 0.03694814071059227, 0.044192347675561905, 0.011306974105536938, -0.005619309842586517, -0.04506392031908035, -0.042298514395952225, -0.014506101608276367, -0.0036004597786813974, -0.06325151771306992, ...
Learning to Learn without Gradient Descent by Gradient Descent
https://proceedings.mlr.press/v70/chen17e.html
[ "Yutian Chen", "Matthew W. Hoffman", "Sergio Gómez Colmenarejo", "Misha Denil", "Timothy P. Lillicrap", "Matt Botvinick", "Nando Freitas" ]
null
null
We learn recurrent neural network optimizers trained on simple synthetic functions by gradient descent. We show that these learned optimizers exhibit a remarkable degree of transfer in that they can be used to efficiently optimize a broad range of derivative-free black-box functions, including Gaussian process bandits,...
[]
null
78
1611.03824
title_snapshot
[ -0.01787693053483963, -0.015031329356133938, 0.007310508284717798, 0.03404269367456436, 0.03970661759376526, 0.04142925515770912, 0.02237573079764843, 0.007958327420055866, -0.022466029971837997, -0.015059489756822586, -0.01915566809475422, 0.04129710793495178, -0.056913405656814575, -0.00...
Identification and Model Testing in Linear Structural Equation Models using Auxiliary Variables
https://proceedings.mlr.press/v70/chen17f.html
[ "Bryant Chen", "Daniel Kumor", "Elias Bareinboim" ]
null
null
We developed a novel approach to identification and model testing in linear structural equation models (SEMs) based on auxiliary variables (AVs), which generalizes a widely-used family of methods known as instrumental variables. The identification problem is concerned with the conditions under which causal parameters c...
[]
null
79
1612.03451
title_snapshot
[ -0.00830270629376173, 0.01603236049413681, -0.035773903131484985, 0.010097851976752281, 0.02076200395822525, 0.0774090364575386, 0.07203678786754608, 0.007918396964669228, -0.02135378122329712, -0.03506629168987274, 0.010389802046120167, -0.0006277650245465338, -0.037130240350961685, 0.001...
Toward Efficient and Accurate Covariance Matrix Estimation on Compressed Data
https://proceedings.mlr.press/v70/chen17g.html
[ "Xixian Chen", "Michael R. Lyu", "Irwin King" ]
null
null
Estimating covariance matrices is a fundamental technique in various domains, most notably in machine learning and signal processing. To tackle the challenges of extensive communication costs, large storage capacity requirements, and high processing time complexity when handling massive high-dimensional and distributed...
[]
null
80
null
null
[ -0.013029142282903194, 0.00025814230320975184, 0.007993129082024097, 0.002974641742184758, 0.035292595624923706, 0.044012513011693954, 0.02854299731552601, -0.017665142193436623, -0.009503317065536976, -0.052716806530952454, -0.015397781506180763, -0.02500576339662075, -0.06362198293209076, ...
Online Partial Least Square Optimization: Dropping Convexity for Better Efficiency and Scalability
https://proceedings.mlr.press/v70/chen17h.html
[ "Zhehui Chen", "Lin F. Yang", "Chris Junchi Li", "Tuo Zhao" ]
null
null
Multiview representation learning is popular for latent factor analysis. Many existing approaches formulate the multiview representation learning as convex optimization problems, where global optima can be obtained by certain algorithms in polynomial time. However, many evidences have corroborated that heuristic noncon...
[]
null
81
null
null
[ -0.02455427311360836, -0.004680695477873087, 0.016439298167824745, 0.031865596771240234, 0.023246560245752335, 0.05734570324420929, 0.014265473000705242, -0.0009028154308907688, -0.008690414018929005, -0.049213260412216187, -0.01812531054019928, -0.010111453011631966, -0.05532773211598396, ...
Learning to Aggregate Ordinal Labels by Maximizing Separating Width
https://proceedings.mlr.press/v70/chen17i.html
[ "Guangyong Chen", "Shengyu Zhang", "Di Lin", "Hui Huang", "Pheng Ann Heng" ]
null
null
While crowdsourcing has been a cost and time efficient method to label massive samples, one critical issue is quality control, for which the key challenge is to infer the ground truth from noisy or even adversarial data by various users. A large class of crowdsourcing problems, such as those involving age, grade, level...
[]
null
82
null
null
[ 0.011149926111102104, -0.005766007117927074, 0.006494979374110699, 0.028980668634176254, 0.0063878619112074375, 0.02704920992255211, 0.021423356607556343, -0.022836966440081596, -0.03901518136262894, -0.02083081379532814, 0.007049175910651684, 0.00200735405087471, -0.07397264242172241, -0....
Nearly Optimal Robust Matrix Completion
https://proceedings.mlr.press/v70/cherapanamjeri17a.html
[ "Yeshwanth Cherapanamjeri", "Kartik Gupta", "Prateek Jain" ]
null
null
In this paper, we consider the problem of Robust Matrix Completion (RMC) where the goal is to recover a low-rank matrix by observing a small number of its entries out of which a few can be arbitrarily corrupted. We propose a simple projected gradient descent-based method to estimate the low-rank matrix that alternately...
[]
null
83
1606.07315
title_snapshot
[ -0.013340773060917854, -0.01990407705307007, 0.03524673730134964, 0.039401158690452576, 0.030966194346547127, 0.0179082490503788, 0.00798171665519476, -0.0038146069273352623, -0.042252399027347565, -0.04268956556916237, -0.03424602001905441, -0.013010745868086815, -0.04063110053539276, -0....
Algorithms for $\ell_p$ Low-Rank Approximation
https://proceedings.mlr.press/v70/chierichetti17a.html
[ "Flavio Chierichetti", "Sreenivas Gollapudi", "Ravi Kumar", "Silvio Lattanzi", "Rina Panigrahy", "David P. Woodruff" ]
null
null
We consider the problem of approximating a given matrix by a low-rank matrix so as to minimize the entrywise $\ell_p$-approximation error, for any $p \geq 1$; the case $p = 2$ is the classical SVD problem. We obtain the first provably good approximation algorithms for this robust version of low-rank approximation that ...
[]
null
84
1705.06730
title_snapshot
[ -0.05258918181061745, -0.002908322960138321, 0.03457760810852051, 0.018536431714892387, 0.022475026547908783, 0.022760950028896332, 0.017754731699824333, -0.0248978640884161, -0.03671189025044441, -0.04041191190481186, -0.014876032248139381, -0.013873069547116756, -0.04383048415184021, -0....
MEC: Memory-efficient Convolution for Deep Neural Network
https://proceedings.mlr.press/v70/cho17a.html
[ "Minsik Cho", "Daniel Brand" ]
null
null
Convolution is a critical component in modern deep neural networks, thus several algorithms for convolution have been developed. Direct convolution is simple but suffers from poor performance. As an alternative, multiple indirect methods have been proposed including im2col-based convolution, FFT-based convolution, or W...
[]
null
85
1706.06873
title_snapshot
[ -0.007256723940372467, -0.031129375100135803, -0.008860220201313496, 0.019158760085701942, 0.03258407488465309, 0.05666318163275719, 0.008961526677012444, 0.007574216928333044, -0.023376720026135445, -0.03191468119621277, 0.012956968508660793, -0.014810820110142231, -0.05113564431667328, 0...
On Relaxing Determinism in Arithmetic Circuits
https://proceedings.mlr.press/v70/choi17a.html
[ "Arthur Choi", "Adnan Darwiche" ]
null
null
The past decade has seen a significant interest in learning tractable probabilistic representations. Arithmetic circuits (ACs) were among the first proposed tractable representations, with some subsequent representations being instances of ACs with weaker or stronger properties. In this paper, we provide a formal basis...
[]
null
86
1708.06846
title_snapshot
[ -0.03127743676304817, 0.015978431329131126, -0.016842270269989967, 0.04769717529416084, 0.03882948309183121, 0.03241795673966408, 0.02081308327615261, 0.006764681544154882, -0.038350269198417664, -0.01971450075507164, 0.008784238249063492, -0.012310627847909927, -0.0636385977268219, 0.0152...
Improving Stochastic Policy Gradients in Continuous Control with Deep Reinforcement Learning using the Beta Distribution
https://proceedings.mlr.press/v70/chou17a.html
[ "Po-Wei Chou", "Daniel Maturana", "Sebastian Scherer" ]
null
null
Recently, reinforcement learning with deep neural networks has achieved great success in challenging continuous control problems such as 3D locomotion and robotic manipulation. However, in real-world control problems, the actions one can take are bounded by physical constraints, which introduces a bias when the standar...
[]
null
87
null
null
[ 0.004825318697839975, -0.0244480911642313, -0.021399229764938354, 0.04023733362555504, 0.03321829065680504, 0.04803139343857765, 0.03876151144504547, -0.00797149259597063, -0.01867908425629139, -0.036667678505182266, 0.010954099707305431, -0.008369240909814835, -0.10666725784540176, 0.0071...
On Kernelized Multi-armed Bandits
https://proceedings.mlr.press/v70/chowdhury17a.html
[ "Sayak Ray Chowdhury", "Aditya Gopalan" ]
null
null
We consider the stochastic bandit problem with a continuous set of arms, with the expected reward function over the arms assumed to be fixed but unknown. We provide two new Gaussian process-based algorithms for continuous bandit optimization – Improved GP-UCB (IGP-UCB) and GP-Thomson sampling (GP-TS), and derive corres...
[]
null
88
1704.00445
title_snapshot
[ -0.018561704084277153, -0.014612462371587753, 0.022980879992246628, 0.061559341847896576, 0.019469834864139557, 0.04488895833492279, 0.030595563352108, 0.0033689371775835752, -0.0066382996737957, -0.047756463289260864, -0.025662001222372055, 0.013351177796721458, -0.056173909455537796, -0....
Parseval Networks: Improving Robustness to Adversarial Examples
https://proceedings.mlr.press/v70/cisse17a.html
[ "Moustapha Cisse", "Piotr Bojanowski", "Edouard Grave", "Yann Dauphin", "Nicolas Usunier" ]
null
null
We introduce Parseval networks, a form of deep neural networks in which the Lipschitz constant of linear, convolutional and aggregation layers is constrained to be smaller than $1$. Parseval networks are empirically and theoretically motivated by an analysis of the robustness of the predictions made by deep neural netw...
[]
null
89
1704.08847
title_snapshot
[ -0.014795084483921528, -0.008107298985123634, -0.0169974397867918, 0.06170385703444481, 0.0133692417293787, 0.03514225035905838, 0.015543358400464058, -0.006162924692034721, 0.005218100268393755, -0.04007028788328171, -0.004524194169789553, -0.023052331060171127, -0.056850288063287735, 0.0...
Deep Latent Dirichlet Allocation with Topic-Layer-Adaptive Stochastic Gradient Riemannian MCMC
https://proceedings.mlr.press/v70/cong17a.html
[ "Yulai Cong", "Bo Chen", "Hongwei Liu", "Mingyuan Zhou" ]
null
null
It is challenging to develop stochastic gradient based scalable inference for deep discrete latent variable models (LVMs), due to the difficulties in not only computing the gradients, but also adapting the step sizes to different latent factors and hidden layers. For the Poisson gamma belief network (PGBN), a recently ...
[]
null
90
1706.01724
title_snapshot
[ -0.01133765745908022, -0.03598112612962723, 0.008348600938916206, 0.05718890205025673, 0.023993240669369698, 0.04661719873547554, 0.016213731840252876, 0.0031705608125776052, -0.012957977131009102, -0.026943225413560867, -0.013158811256289482, 0.005052138585597277, -0.04258772358298302, 0....
AdaNet: Adaptive Structural Learning of Artificial Neural Networks
https://proceedings.mlr.press/v70/cortes17a.html
[ "Corinna Cortes", "Xavier Gonzalvo", "Vitaly Kuznetsov", "Mehryar Mohri", "Scott Yang" ]
null
null
We present a new framework for analyzing and learning artificial neural networks. Our approach simultaneously and adaptively learns both the structure of the network as well as its weights. The methodology is based upon and accompanied by strong data-dependent theoretical learning guarantees, so that the final network ...
[]
null
91
1607.01097
title_snapshot
[ -0.02788693644106388, -0.008530797436833382, -0.027863070368766785, 0.021389709785580635, 0.036405667662620544, 0.04669170826673508, 0.014258403331041336, -0.01471742708235979, -0.038916002959012985, -0.02018880285322666, 0.021761579439044, 0.01594081148505211, -0.05326038599014282, -0.010...
Random Feature Expansions for Deep Gaussian Processes
https://proceedings.mlr.press/v70/cutajar17a.html
[ "Kurt Cutajar", "Edwin V. Bonilla", "Pietro Michiardi", "Maurizio Filippone" ]
null
null
The composition of multiple Gaussian Processes as a Deep Gaussian Process DGP enables a deep probabilistic nonparametric approach to flexibly tackle complex machine learning problems with sound quantification of uncertainty. Existing inference approaches for DGP models have limited scalability and are notoriously cumbe...
[]
null
92
1610.04386
title_snapshot
[ 0.012645391747355461, -0.008849727921187878, -0.009040728211402893, 0.04511389881372452, 0.01909908838570118, 0.06452310085296631, 0.024726679548621178, -0.0124495979398489, -0.01177326962351799, -0.05462150275707245, -0.023462360724806786, -0.011335022747516632, -0.07763675600290298, 0.00...
Soft-DTW: a Differentiable Loss Function for Time-Series
https://proceedings.mlr.press/v70/cuturi17a.html
[ "Marco Cuturi", "Mathieu Blondel" ]
null
null
We propose in this paper a differentiable learning loss between time series, building upon the celebrated dynamic time warping (DTW) discrepancy. Unlike the Euclidean distance, DTW can compare time series of variable size and is robust to shifts or dilatations across the time dimension. To compute DTW, one typically so...
[]
null
93
1703.01541
title_snapshot
[ -0.010370069183409214, -0.040494147688150406, 0.0019445591606199741, 0.04150060564279556, 0.028086306527256966, 0.050680238753557205, 0.030221981927752495, 0.017363019287586212, -0.019460266456007957, -0.054999422281980515, 0.01487002708017826, 0.009096750989556313, -0.05067792907357216, 0...
Understanding Synthetic Gradients and Decoupled Neural Interfaces
https://proceedings.mlr.press/v70/czarnecki17a.html
[ "Wojciech Marian Czarnecki", "Grzegorz Świrszcz", "Max Jaderberg", "Simon Osindero", "Oriol Vinyals", "Koray Kavukcuoglu" ]
null
null
When training neural networks, the use of Synthetic Gradients (SG) allows layers or modules to be trained without update locking – without waiting for a true error gradient to be backpropagated – resulting in Decoupled Neural Interfaces (DNIs). This unlocked ability of being able to update parts of a neural network asy...
[]
null
94
1703.00522
title_snapshot
[ -0.03622714430093765, -0.014388678595423698, -0.012299803085625172, 0.0343567356467247, 0.04595814645290375, 0.05020900443196297, 0.01603475771844387, 0.006047565955668688, -0.021795714274048805, -0.029660610482096672, 0.01230465155094862, 0.010460878722369671, -0.06822965294122696, -0.000...
Stochastic Generative Hashing
https://proceedings.mlr.press/v70/dai17a.html
[ "Bo Dai", "Ruiqi Guo", "Sanjiv Kumar", "Niao He", "Le Song" ]
null
null
Learning-based binary hashing has become a powerful paradigm for fast search and retrieval in massive databases. However, due to the requirement of discrete outputs for the hash functions, learning such functions is known to be very challenging. In addition, the objective functions adopted by existing hashing technique...
[]
null
95
1701.02815
title_snapshot
[ -0.012584018521010876, -0.01395905390381813, -0.019066771492362022, 0.0851278007030487, 0.041003014892339706, 0.025513986125588417, 0.017129946500062943, 0.011978712864220142, -0.01686224713921547, -0.022138303145766258, -0.02715597115457058, -0.03158219903707504, -0.05929179862141609, 0.0...
Logarithmic Time One-Against-Some
https://proceedings.mlr.press/v70/daume17a.html
[ "Hal Daumé III", "Nikos Karampatziakis", "John Langford", "Paul Mineiro" ]
null
null
We create a new online reduction of multiclass classification to binary classification for which training and prediction time scale logarithmically with the number of classes. We show that several simple techniques give rise to an algorithm which is superior to previous logarithmic time classification approaches while ...
[]
null
96
1606.04988
title_snapshot
[ -0.004766459111124277, -0.011489896103739738, 0.01176463719457388, 0.016803335398435593, 0.040953658521175385, 0.02800891548395157, 0.018946506083011627, -0.007342022843658924, -0.024217994883656502, -0.037050869315862656, 0.03748909756541252, -0.00047156729851849377, -0.060181841254234314, ...
Language Modeling with Gated Convolutional Networks
https://proceedings.mlr.press/v70/dauphin17a.html
[ "Yann N. Dauphin", "Angela Fan", "Michael Auli", "David Grangier" ]
null
null
The pre-dominant approach to language modeling to date is based on recurrent neural networks. Their success on this task is often linked to their ability to capture unbounded context. In this paper we develop a finite context approach through stacked convolutions, which can be more efficient since they allow paralleliz...
[]
null
97
1612.08083
title_snapshot
[ -0.02193489670753479, -0.021878089755773544, -0.01854912005364895, 0.03728393092751503, 0.01343053113669157, 0.04021817073225975, 0.02882627211511135, 0.03185870498418808, 0.008178838528692722, -0.00907522439956665, 0.0014828097773715854, 0.008212411776185036, -0.057291604578495026, 0.0126...
An Infinite Hidden Markov Model With Similarity-Biased Transitions
https://proceedings.mlr.press/v70/dawson17a.html
[ "Colin Reimer Dawson", "Chaofan Huang", "Clayton T. Morrison" ]
null
null
We describe a generalization of the Hierarchical Dirichlet Process Hidden Markov Model (HDP-HMM) which is able to encode prior information that state transitions are more likely between “nearby” states. This is accomplished by defining a similarity function on the state space and scaling transition probabilities by pai...
[]
null
98
1707.06756
title_snapshot
[ -0.02927614003419876, 0.005389212630689144, -0.005815752316266298, 0.02815263345837593, 0.04439829662442207, 0.040313683450222015, 0.05528690293431282, 0.030242804437875748, -0.015498858876526356, -0.049088675528764725, -0.012908720411360264, 0.002170548075810075, -0.07034137099981308, -0....
Distributed Batch Gaussian Process Optimization
https://proceedings.mlr.press/v70/daxberger17a.html
[ "Erik A. Daxberger", "Bryan Kian Hsiang Low" ]
null
null
This paper presents a novel distributed batch Gaussian process upper confidence bound (DB-GP-UCB) algorithm for performing batch Bayesian optimization (BO) of highly complex, costly-to-evaluate black-box objective functions. In contrast to existing batch BO algorithms, DB-GP-UCB can jointly optimize a batch of inputs (...
[]
null
99
null
null
[ -0.038948606699705124, -0.012551027350127697, -0.011127308942377567, 0.04236915707588196, 0.02414991706609726, 0.05336644873023033, 0.029517782852053642, 0.01217136811465025, -0.0058580441400408745, -0.034177932888269424, -0.015536252409219742, 0.0044591668993234634, -0.08014131337404251, ...
Consistency Analysis for Binary Classification Revisited
https://proceedings.mlr.press/v70/dembczynski17a.html
[ "Krzysztof Dembczyński", "Wojciech Kotłowski", "Oluwasanmi Koyejo", "Nagarajan Natarajan" ]
null
null
Statistical learning theory is at an inflection point enabled by recent advances in understanding and optimizing a wide range of metrics. Of particular interest are non-decomposable metrics such as the F-measure and the Jaccard measure which cannot be represented as a simple average over examples. Non-decomposability i...
[]
null
100
null
null
[ -0.010225758887827396, -0.02774231880903244, -0.025021323934197426, 0.030677417293190956, 0.05387149751186371, 0.027914971113204956, 0.02343670278787613, 0.004221368581056595, -0.02267085574567318, -0.03638995811343193, -0.04762379080057144, -0.0016024343203753233, -0.06398270279169083, -0...