Non-asymptotic error bounds for constant stepsize stochastic approximation for tracking mobile agents

被引：2
|
作者：
Kumar, Bhumesh [1 ,2 ]
Borkar, Vivek [1 ]
Shetty, Akhil [1 ,3 ]
机构：
[1] Indian Inst Technol, Dept Elect Engn, Mumbai 400076, Maharashtra, India
[2] Univ Wisconsin, Dept Elect & Comp Engn, 1415 Johnson Dr, Madison, WI 53706 USA
[3] Univ Calif Berkeley, Dept Elect Engn & Comp Sci, Cory Hall,Hearst Ave, Berkeley, CA 94720 USA
来源：
MATHEMATICS OF CONTROL SIGNALS AND SYSTEMS | 2019年 / 31卷 / 04期
关键词：
Stochastic approximation; Constant stepsize; Non-asymptotic bound; Alekseev's formula; Martingale concentration inequalities; Perturbation analysis; Non-stationary optimization; EXPONENTIAL STABILITY; ASYMPTOTIC ANALYSIS; ALGORITHMS; SYSTEMS; CONVERGENCE; SIZE;
D O I：
10.1007/s00498-019-00249-4
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
This work revisits the constant stepsize stochastic approximation algorithm for tracking a slowly moving target and obtains a bound for the tracking error that is valid for the entire time axis, using the Alekseev nonlinear variation of constants formula. It is the first non-asymptotic bound for the entire time axis in the sense that it is not based on the vanishing stepsize limit and associated limit theorems unlike prior works, and captures clearly the dependence on problem parameters and the dimension.
引用
收藏
页码：589 / 614
页数：26
相关论文
共 50 条
  • [21] Normal Approximation for Stochastic Gradient Descent via Non-Asymptotic Rates of Martingale CLT
    Anastasiou, Andreas
    Balasubramanian, Krishnakumar
    Erdogdu, Murat A.
    CONFERENCE ON LEARNING THEORY, VOL 99, 2019, 99
  • [22] Linear Model Regression on Time-series Data: Non-asymptotic Error Bounds and Applications
    Alaeddini, Atiye
    Alemzadeh, Siavash
    Mesbahi, Afshin
    Mesbahi, Mehran
    2018 IEEE CONFERENCE ON DECISION AND CONTROL (CDC), 2018, : 2259 - 2264
  • [23] Multiscale Dictionary Learning: Non-Asymptotic Bounds and Robustness
    Maggioni, Mauro
    Minsker, Stanislav
    Strawn, Nate
    JOURNAL OF MACHINE LEARNING RESEARCH, 2016, 17
  • [24] Non-Asymptotic Achievability Bounds in Multiuser Information Theory
    Verdue, Sergio
    2012 50TH ANNUAL ALLERTON CONFERENCE ON COMMUNICATION, CONTROL, AND COMPUTING (ALLERTON), 2012, : 1 - 8
  • [25] Tight High Probability Bounds for Linear Stochastic Approximation with Fixed Stepsize
    Durmus, Alain
    Moulines, Eric
    Naumov, Alexey
    Samsonov, Sergey
    Scaman, Kevin
    Wai, Hoi-To
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 34 (NEURIPS 2021), 2021,
  • [26] Recursive Quantile Estimation: Non-Asymptotic Confidence Bounds
    Chen, Likai
    Keilbar, Georg
    Wu, Wei Biao
    JOURNAL OF MACHINE LEARNING RESEARCH, 2023, 24
  • [27] On Linear Stochastic Approximation: Fine-grained Polyak-Ruppert and Non-Asymptotic Concentration
    Mou, Wenlong
    Li, Chris Junchi
    Wainwright, Martin J.
    Bartlett, Peter L.
    Jordan, Michael I.
    CONFERENCE ON LEARNING THEORY, VOL 125, 2020, 125
  • [28] Non-Asymptotic Guarantees for Sampling by Stochastic Gradient Descent
    A. G. Karagulyan
    Journal of Contemporary Mathematical Analysis (Armenian Academy of Sciences), 2019, 54 : 71 - 78
  • [29] On the non-asymptotic and sharp lower tail bounds of random variables
    Zhang, Anru R.
    Zhou, Yuchen
    STAT, 2020, 9 (01):
  • [30] Non-Asymptotic Bounds of AIPW Estimators for Means with Missingness at Random
    Wang, Fei
    Deng, Yuhao
    MATHEMATICS, 2023, 11 (04)
12345