Efficient Large-Scale Gaussian Process Bandits by Believing only Informative Actions

被引:0
|
作者
Bedi, Amrit Singh [1 ]
Peddireddy, Dheeraj [2 ]
Aggarwal, Vaneet [3 ]
Koppel, Alec [1 ]
机构
[1] US Army Res Lab, CISD, 2800 Powder Mill Rd, Adelphi, MD 20783 USA
[2] Purdue Univ, Sch IE, 315 N Grant St, W Lafayette, IN 47907 USA
[3] Purdue Univ, Sch IE & ECE, 315 N Grant St, W Lafayette, IN 47907 USA
来源
LEARNING FOR DYNAMICS AND CONTROL, VOL 120 | 2020年 / 120卷
关键词
multi-armed bandits; Bayesian optimization; Gaussian Processes; adaptive control; OPTIMIZATION; DESIGN;
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
In this work, we cast Bayesian optimization as a multi-armed bandit problem, where the payoff function is sampled from a Gaussian process (GP). Further, we focus on action selections via the GP upper confidence bound (UCB). While numerous prior works use GPs in bandit settings, they do not apply to settings where the total number of iterations T may be large-scale, as the complexity of computing the posterior parameters scales cubically with the number of past observations. To circumvent this computational burden, we propose a simple statistical test: only incorporate an action into the GP posterior when its conditional entropy exceeds an E threshold. Doing so permits us to derive sublinear regret bounds of GP bandit algorithms up to factors depending on the compression parameter E for both discrete and continuous action sets. Moreover, the complexity of the GP posterior remains provably finite. Experimentally, we observe state of the art accuracy and complexity tradeoffs for GP bandit algorithms on various hyper-parameter tuning tasks, suggesting the merits of managing the complexity of GPs in bandit settings.
引用
收藏
页码:924 / 934
页数:11
相关论文
共 50 条
  • [1] Asynchronous Parallel Large-Scale Gaussian Process Regression
    Dang, Zhiyuan
    Gu, Bin
    Deng, Cheng
    Huang, Heng
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2024, 35 (06) : 8683 - 8694
  • [2] A Scalable Gaussian Process for Large-Scale Periodic Data
    Li, Yongxiang
    Pu, Yuting
    Cheng, Changming
    Xiao, Qian
    TECHNOMETRICS, 2023, 65 (03) : 363 - 374
  • [3] Local Gaussian Process Model for Large-Scale Dynamic Computer Experiments
    Zhang, Ru
    Lin, C. Devon
    Ranjan, Pritam
    JOURNAL OF COMPUTATIONAL AND GRAPHICAL STATISTICS, 2018, 27 (04) : 798 - 807
  • [4] Large-scale Gaussian process classification using random decision forests
    B. Fröhlich
    E. Rodner
    M. Kemmler
    J. Denzler
    Pattern Recognition and Image Analysis, 2012, 22 (1) : 113 - 120
  • [5] Bayesian estimation of large-scale simulation models with Gaussian process regression surrogates
    Barde, Sylvain
    COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2024, 196
  • [6] A Global-Local Approximation Framework for Large-Scale Gaussian Process Modeling
    Vakayil, Akhil
    Joseph, V. Roshan
    TECHNOMETRICS, 2024, 66 (02) : 295 - 305
  • [7] Efficient and effective large-scale vaccine distribution
    Muckstadt, John A.
    Klein, Michael G.
    Jackson, Peter L.
    Gougelet, Robert M.
    Hupert, Nathaniel
    INTERNATIONAL JOURNAL OF PRODUCTION ECONOMICS, 2023, 262
  • [8] Integrated process synthesis of large-scale chemical processes
    Krajnc, D
    Pintaric, ZN
    Glavic, P
    EUROPEAN SYMPOSIUM ON COMPUTER-AIDED PROCESS ENGINEERING - 14, 2004, 18 : 229 - 234
  • [9] Online Stochastic Variational Gaussian Process Mapping for Large-Scale Bathymetric SLAM in Real Time
    Torroba, Ignacio
    Cella, Marco
    Teran, Aldo
    Rolleberg, Niklas
    Folkesson, John
    IEEE ROBOTICS AND AUTOMATION LETTERS, 2023, 8 (06): : 3150 - 3157
  • [10] Large-Scale Gaussian Process Inference with Generalized Histogram Intersection Kernels for Visual Recognition Tasks
    Rodner, Erik
    Freytag, Alexander
    Bodesheim, Paul
    Froehlich, Bjoern
    Denzler, Joachim
    INTERNATIONAL JOURNAL OF COMPUTER VISION, 2017, 121 (02) : 253 - 280