999 Hits in 4.8 sec

GRMR: Generalized Regret-Minimizing Representatives [article]

Yanhao Wang, Michael Mathioudakis, Yuchen Li, Kian-Lee Tan
2020 arXiv   pre-print
The regret-minimizing set (RMS) problem is recently proposed for representative discovery from databases.  ...  For two-dimensional databases, we propose an optimal algorithm for GRMR via a transformation into the shortest cycle problem in a directed graph.  ...  On the other hand, both ε-K and H S run faster when ε is larger because of smaller sample sizes for computation.  ... 
arXiv:2007.09634v1 fatcat:oh2nswugbfd4hpj3f6udauvja4

Online Submodular Maximization under a Matroid Constraint with Application to Learning Assignments [article]

Daniel Golovin, Andreas Krause, Matthew Streeter
2014 arXiv   pre-print
We present an efficient algorithm for this general problem and analyze it in the no-regret model.  ...  We show that these and other problems can be formalized as repeatedly selecting an assignment of items to positions to maximize a sequence of monotone submodular functions that arrive one by one.  ...  grant IIS-0953413, and by ONR grant N00014-09-1-1044.  ... 
arXiv:1407.1082v1 fatcat:4js4omc36zfmhbseizd5lvw6y4

Rejection sampling from shape-constrained distributions in sublinear time [article]

Sinho Chewi, Patrik Gerber, Chen Lu, Thibaut Le Gouic, Philippe Rigollet
2021 arXiv   pre-print
The classical algorithm for this task is rejection sampling, and although it has been used in practice for decades, there is surprisingly little study of its fundamental limitations.  ...  Our results provide new algorithms for sampling whose complexity scales sublinearly with the alphabet size.  ...  Philippe Rigollet was supported by NSF awards IIS-1838071, DMS-1712596, and DMS-2022448.  ... 
arXiv:2105.14166v1 fatcat:q66vhdtnxjciba6ia46zwaikqi

Finding and Certifying (Near-)Optimal Strategies in Black-Box Extensive-Form Games [article]

Brian Hu Zhang, Tuomas Sandholm
2021 arXiv   pre-print
However, that work assumed that the black box could sample or expand arbitrary nodes of the game tree at any time, and that a series of exact game solves (via, for example, linear programming) can be conducted  ...  As a bonus, we obtain an equilibrium-finding algorithm with Õ(1/√(T)) convergence rate in the extensive-form game setting that does not rely on a sampling strategy with lower-bounded reach probabilities  ...  Acknowledgements This material is based on work supported by the National Science Foundation under grants IIS-1718457, IIS-1901403, and CCF-1733556, and the ARO under award W911NF2010081.  ... 
arXiv:2009.07384v3 fatcat:lfn4vztrd5dexlh3rqxbxmiu2u

Retaining Data from Streams of Social Platforms with Minimal Regret

Nguyen Thanh Tam, Matthias Weidlich, Duong Chi Thang, Hongzhi Yin, Nguyen Quoc Viet Hung
2017 Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence  
In this paper, we propose techniques to effectively decide which data to retain, such that the induced loss of information, the regret of neglecting certain data, is minimized.  ...  The resulting data streams exceed any reasonable limit for permanent storage, especially since data is often redundant, overlapping, sparse, and generally of low value.  ...  Setting r ∈ (0.5, 1] guarantees convergence [17] , while larger values often lead to higher learning quality and faster convergence (but not monotonically).  ... 
doi:10.24963/ijcai.2017/397 dblp:conf/ijcai/TamWTYH17 fatcat:hbp3msub2zfjjektywma4svcpe

Accelerating Experimental Design by Incorporating Experimenter Hunches [article]

Cheng Li, Santu Rana, Sunil Gupta, Vu Nguyen, Svetha Venkatesh, Alessandra Sutti, David Rubin, Teo Slezak, Murray Height, Mazher Mohammed, Ian Gibson
2019 arXiv   pre-print
For example, sweetness of a candy is monotonic to the sugar content.  ...  This is achieved by a two-stage Gaussian process modeling, where the first stage uses the monotonicity trend to model the underlying property, and the second stage uses 'virtual' samples, sampled from  ...  Thus we make full use of the monotonicity of f (x) and transfer this critical knowledge to g(x) through a set of sampled points.  ... 
arXiv:1907.09065v1 fatcat:j6iuddrq4rdnhoorjc6sb5wzkm

Online Learning for Active Cache Synchronization [article]

Andrey Kolobov, Sébastien Bubeck, Julian Zimmert
2020 arXiv   pre-print
We present MirrorSync, an online learning algorithm for synchronization bandits, establish an adversarial regret of O(T^2/3) for it, and show how to make it practical.  ...  Existing multi-armed bandit (MAB) models make two implicit assumptions: an arm generates a payoff only when it is played, and the agent observes every payoff that is generated.  ...  We would like to thank Nicole Immorlica (Microsoft Research) and the anonymous reviewers for their helpful comments and suggestions regarding this work.  ... 
arXiv:2002.12014v2 fatcat:msxdj3lopnbs7f7osbxzy4cz3i

A Novel Spatio-Temporal Multi-Task Approach for the Prediction of Diabetes-Related Complication: a Cardiopathy Case of Study

Luca Romeo, Giuseppe Armentano, Antonio Nicolucci, Marco Vespasiani, Giacomo Vespasiani, Emanuele Frontoni
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
a regularization term and a graph-based approach (i.e., encoding the task relatedness using the structure matrix).  ...  Experimental results on a real-world EHR dataset demonstrate the robust performance and the interpretability of the proposed approach.  ...  The corresponding authors Yong Yu and Weinan Zhang thank the support of NSFC 61702327 and 61772333. Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  ... 
doi:10.24963/ijcai.2020/588 dblp:conf/ijcai/ChenL0YL20 fatcat:sqcqcgwd2vfixhniglzuejqdd4

Algorithms and Learning for Fair Portfolio Design [article]

Emily Diana, Travis Dick, Hadi Elzayn, Michael Kearns, Aaron Roth, Zachary Schutzman, Saeed Sharifi-Malvajerdi, Juba Ziani
2020 arXiv   pre-print
Our main results are algorithms for optimal and near-optimal portfolio design for both social welfare and fairness objectives, both with and without assumptions on the underlying group structure.  ...  We describe an efficient algorithm based on an internal two-player zero-sum game that learns near-optimal fair portfolios ex ante and show experimentally that it can be used to obtain a small set of fair  ...  D Approximate Population Regret Minimization via Greedy Algorithm Recall that given S = {τ i } n i=1 and set c ⊆ S of products, the population regret of S is given by R S (c) = 1 n n i=1 R τi (c) = 1 n  ... 
arXiv:2006.07281v1 fatcat:fhr7srflsjfubm4jw5uee4fijy

Online Ranking with Top-1 Feedback [article]

Sougata Chaudhuri, Ambuj Tewari
2016 arXiv   pre-print
We provide a comprehensive set of results regarding learnability under this challenging setting. For PairwiseLoss and DCG, two popular ranking measures, we prove that the minimax regret is Θ(T^2/3).  ...  Moreover, the minimax regret is achievable using an efficient strategy that only spends O(m m) time per round. The same efficient strategy achieves O(T^2/3) regret for Precision@k.  ...  In Algorithm 1, since the average of the relevance vectors per block was estimated by uniform sampling according to Lemma 8, the algorithm was run 10 times, with the same set of relevance vectors, for  ... 
arXiv:1410.1103v3 fatcat:zt4yz24svfhrdhxbfoynhc2sjy

Lenient Regret and Good-Action Identification in Gaussian Process Bandits [article]

Xu Cai, Selwyn Gomes, Jonathan Scarlett
2021 arXiv   pre-print
On the theoretical side, we study various lenient regret notions in which all near-optimal actions incur zero penalty, and provide upper bounds on the lenient regret for GP-UCB and an elimination algorithm  ...  We experimentally find that such algorithms can often find a good action faster than standard optimization-based approaches.  ...  set of K samples of maximum function values upon choosing x, which can be generated in an identical manner to max-value entropy search (MES) via a Gumbel distribution approximation (Wang & Jegelka, 2017  ... 
arXiv:2102.05793v2 fatcat:srlubdetzzdvrdqtucqgmqguou

Discovering Valuable items from Massive Data

Hastagiri P. Vanchinathan, Andreas Marfurt, Charles-Antoine Robelin, Donald Kossmann, Andreas Krause
2015 Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD '15  
We extend GP-Select to be able to discover sets that simultaneously have high utility and are diverse.  ...  Our preference for diversity can be specified as an arbitrary monotone submodular function that quantifies the diminishing returns obtained when selecting similar items.  ...  This research was supported in part by SNSF grant 200020 159557, ERC StG 307036 and a Microsoft Research Faculty Fellowship. The authors wish to thank Christian Widmer for providing the MHC data.  ... 
doi:10.1145/2783258.2783360 dblp:conf/kdd/VanchinathanMRK15 fatcat:jiprn43dl5capbzo2oojgeu7wi

Exploiting correlation and budget constraints in Bayesian multi-armed bandit optimization [article]

Matthew W. Hoffman, Bobak Shahriari, Nando de Freitas
2013 arXiv   pre-print
We introduce a Bayesian approach for this problem and show that it empirically outperforms both the existing frequentist counterpart and other Bayesian optimization methods.  ...  The paper presents comprehensive comparisons of the proposed approach, Thompson sampling, classical Bayesian optimization techniques, more recent Bayesian bandit approaches, and state-of-the-art best arm  ...  Algorithm 1 BayesGap 1: for t = 1, . . . , T do select arm a t = arg max k∈{j(t),J(t)} s k (t) 5: observe y t ∼ ν at (•) 6: update posterior μkt and σkt 7: update bound on H and re-compute β 8: 2:set J  ... 
arXiv:1303.6746v4 fatcat:ovreptonz5amfbvz7jzw2iwhae

Is Reinforcement Learning More Difficult Than Bandits? A Near-optimal Algorithm Escaping the Curse of Horizon [article]

Zihan Zhang, Xiangyang Ji, Simon S. Du
2021 arXiv   pre-print
In particular, the constants in the bonus should be subtly setting to ensure optimism and monotonicity.  ...  the running time in [Wang et al. 2020] and significantly improves the dependency on S, A and K in sample complexity.  ...  Comparing with the algorithm in Wang et al. (2020), our algorithm is exponentially faster and achieves significantly better sample complexity in terms of S, A, ǫ.  ... 
arXiv:2009.13503v2 fatcat:vrirdr2x6rfubdipvpaws5aesm

The Influence of Shape Constraints on the Thresholding Bandit Problem [article]

James Cheshire, Pierre Menard, Alexandra Carpentier
2021 arXiv   pre-print
In the fixed budget setting, we provide problem independent minimax rates for the expected regret in all settings, as well as associated algorithms.  ...  We prove that the minimax rates for the regret are (i) √(log(K)K/T) for TBP, (ii) √(log(K)/T) for MTBP, (iii) √(K/T) for UTBP and (iv) √(loglog K/T) for CTBP, where K is the number of arms and T is the  ...  Ménard is supported by the European CHISTERA project DELTA, and partially supported by the SFI Sachsen-Anhalt for the project RE-BCI, and by the UFA-DFH through the French-German Doktorandenkolleg CDFA  ... 
arXiv:2006.10006v3 fatcat:6gvkwqhpinevlg3pss4f6e5n6y
« Previous Showing results 1 — 15 out of 999 results