Stochastic Rank-1 Bandits Stochastic Rank-1 Bandits

被引:0
|
作者
Katariya, Sumeet [1 ]
Kveton, Branislav [2 ]
Szepesvari, Csaba [3 ]
Vernade, Claire [4 ]
Wen, Zheng [2 ]
机构
[1] Univ Wisconsin, Dept ECE, Madison, WI 53706 USA
[2] Adobe Res, San Jose, CA USA
[3] Univ Alberta, Dept Comp Sci, Edmonton, AB, Canada
[4] Telecom ParisTech, Paris, France
来源
ARTIFICIAL INTELLIGENCE AND STATISTICS, VOL 54 | 2017年 / 54卷
基金
加拿大自然科学与工程研究理事会;
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We propose stochastic rank-1 bandits, a class of online learning problems where at each step a learning agent chooses a pair of row and column arms, and receives the product of their values as a reward. The main challenge of the problem is that the individual values of the row and column are unobserved. We assume that these values are stochastic and drawn independently. We propose a computationally-efficient algorithm for solving our problem, which we call RanklElim. We derive a O((K + L) (1/Delta) log n) upper bound on its n-step regret, where K is the number of rows, L is the number of columns, and Delta is the minimum of the row and column gaps; under the assumption that the mean row and column rewards are bounded away from zero. To the best of our knowledge, we present the first bandit algorithm that finds the maximum entry of a rank-1 matrix whose regret is linear in K L, 1/Delta, and log n. We also derive a nearly matching lower bound. Finally, we evaluate RanklElim empirically on multiple problems. We observe that it leverages the structure of our problems and can learn nearoptimal solutions even if our modeling assumptions are mildly violated.
引用
收藏
页码:392 / 401
页数:10
相关论文
共 50 条
  • [1] Bernoulli Rank-1 Bandits for Click Feedback
    Katariya, Sumeet
    Kveton, Branislav
    Szepesvari, Csaba
    Vernade, Claire
    Wen, Zheng
    PROCEEDINGS OF THE TWENTY-SIXTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2017, : 2001 - 2007
  • [2] Rank-1 Representers
    A. Eberhard
    D. Ralph
    Journal of Mathematical Sciences, 2003, 115 (5) : 2653 - 2700
  • [3] BOUNDED RANK-1 TRANSFORMATIONS
    Gao, Su
    Hill, Aaron
    JOURNAL D ANALYSE MATHEMATIQUE, 2016, 129 : 341 - 365
  • [4] Implicit Bias of (Stochastic) Gradient Descent for Rank-1 Linear Neural Network
    Lyu, Bochen
    Zhu, Zhanxing
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 36 (NEURIPS 2023), 2023,
  • [5] Simulation on rank-1 lattices
    Dammertz, Holger
    Keller, Alexander
    Dammertz, Sabrina
    MONTE CARLO AND QUASI-MONTE CARLO METHODS 2006, 2008, : 205 - 216
  • [6] Stratification by rank-1 lattices
    Keller, A
    MONTE CARLO AND QUASI-MONTE CARLO METHODS 2002, 2004, : 299 - 313
  • [7] ON FREE EXTENSIONS OF RANK-1
    SANDLER, R
    MATHEMATISCHE ZEITSCHRIFT, 1969, 111 (03) : 233 - &
  • [8] Bounded rank-1 transformations
    Su Gao
    Aaron Hill
    Journal d'Analyse Mathématique, 2016, 129 : 341 - 365
  • [9] Textures on Rank-1 Lattices
    Dammertz, Sabrina
    Dammertz, Holger
    Keller, Alexander
    Lensch, Hendrik P. A.
    COMPUTER GRAPHICS FORUM, 2009, 28 (07) : 1945 - 1954
  • [10] RANK-1 MATRICES A, B, A + B
    GLICK, N
    AMERICAN MATHEMATICAL MONTHLY, 1982, 89 (02): : 133 - 133