Langevin Monte Carlo without smoothness

被引:0
|
作者
Chatterji, Niladri S. [1 ]
Diakonikolas, Jelena [2 ]
Jordan, Michael I. [1 ]
Bartlett, Peter L. [1 ]
机构
[1] Univ Calif Berkeley, Berkeley, CA 94720 USA
[2] UW Madison, Madison, WI USA
关键词
ALGORITHM;
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
Langevin Monte Carlo (LMC) is an iterative algorithm used to generate samples from a distribution that is known only up to a normalizing constant. The nonasymptotic dependence of its mixing time on the dimension and target accuracy is understood mainly in the setting of smooth (gradient-Lipschitz) log-densities, a serious limitation for applications in machine learning. In this paper, we remove this limitation, providing polynomial-time convergence guarantees for a variant of LMC in the setting of nonsmooth log-concave distributions. At a high level, our results follow by leveraging the implicit smoothing of the log-density that comes from a small Gaussian perturbation that we add to the iterates of the algorithm and controlling the bias and variance that are induced by this perturbation.
引用
下载
收藏
页码:1716 / 1725
页数:10
相关论文
共 50 条
  • [1] Fluctuation without dissipation: Microcanonical Langevin Monte Carlo
    Department of Physics, University of California, Berkeley
    CA
    94720, United States
    不详
    CA
    93720, United States
    Proc. Mach. Learn. Res., 1600, (111-126):
  • [2] CONSTRAINED ENSEMBLE LANGEVIN MONTE CARLO
    Ding, Zhiyan
    Li, Qin
    FOUNDATIONS OF DATA SCIENCE, 2022, 4 (01): : 37 - 70
  • [3] Langevin Monte Carlo for Contextual Bandits
    Xu, Pan
    Zheng, Hongkai
    Mazumdar, Eric
    Azizzadenesheli, Kamyar
    Anandkumar, Anima
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 162, 2022,
  • [4] Lévy Langevin Monte Carlo
    Oechsler, David
    STATISTICS AND COMPUTING, 2024, 34 (01)
  • [5] Lévy Langevin Monte Carlo
    David Oechsler
    Statistics and Computing, 2024, 34
  • [6] Langevin Quasi-Monte Carlo
    Liu, Sifan
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 36 (NEURIPS 2023), 2023,
  • [7] Langevin Monte Carlo Filtering for Target Tracking
    Garcia, Fernando J. Iglesias
    Bocquel, Melanie
    Driessen, Hans
    2015 18TH INTERNATIONAL CONFERENCE ON INFORMATION FUSION (FUSION), 2015, : 82 - 89
  • [8] Higher order Langevin Monte Carlo algorithm
    Sabanis, Sotirios
    Zhang, Ying
    ELECTRONIC JOURNAL OF STATISTICS, 2019, 13 (02): : 3805 - 3850
  • [9] Random Coordinate Underdamped Langevin Monte Carlo
    Ding, Zhiyan
    Li, Qin
    Lu, Jianfeng
    Wright, Stephen J.
    24TH INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND STATISTICS (AISTATS), 2021, 130
  • [10] COMPARISON OF LANGEVIN AND MONTE-CARLO DYNAMICS
    ETTELAIE, R
    MOORE, MA
    JOURNAL OF PHYSICS A-MATHEMATICAL AND GENERAL, 1984, 17 (18): : 3505 - 3520