Finite-sample analysis of iterate averaging method for stochastic approximation with quadratic loss function

被引:0
作者
Yang, Kaining [1 ]
Wang, Kunbo [1 ]
Feng, Chen [1 ]
机构
[1] Johns Hopkins Univ, Dept Appl Math & Stat, Baltimore, MD 21218 USA
来源
2017 51ST ANNUAL CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS) | 2017年
关键词
Stochastic Approximation; Iterate Averaging; Finite Sample;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Consider the stochastic approximation algorithms with quadratic loss functions, where we have noisy measurements of gradients at each iteration. A widely-used technique of stochastic approximation is to average some or all of the iterates in order to reduce the variance of the resulting estimate. Under proper settings for coefficients, the averaged sequence converges to its limit at an optimum rate. In practice, however, the results on iterate averaging are more mixed than the above suggests. Under proper assumptions, we provide a formal analysis under finite iterations and derive conditions under which iterate averaging benefits the algorithm output, in terms of reducing mean squared errors. Simulations and examples support the practical importance of the given conditions.
引用
收藏
页数:6
相关论文
共 18 条
[1]   APPROXIMATION METHODS WHICH CONVERGE WITH PROBABILITY ONE [J].
BLUM, JR .
ANNALS OF MATHEMATICAL STATISTICS, 1954, 25 (02) :382-386
[2]   MULTIDIMENSIONAL STOCHASTIC APPROXIMATION METHODS [J].
BLUM, JR .
ANNALS OF MATHEMATICAL STATISTICS, 1954, 25 (04) :737-744
[3]   ON A STOCHASTIC APPROXIMATION METHOD [J].
CHUNG, KL .
ANNALS OF MATHEMATICAL STATISTICS, 1954, 25 (03) :463-483
[4]   STOCHASTIC ESTIMATION OF THE MAXIMUM OF A REGRESSION FUNCTION [J].
KIEFER, J ;
WOLFOWITZ, J .
ANNALS OF MATHEMATICAL STATISTICS, 1952, 23 (03) :462-466
[5]   STOCHASTIC-APPROXIMATION WITH AVERAGING OF THE ITERATES - OPTIMAL ASYMPTOTIC RATE OF CONVERGENCE FOR GENERAL PROCESSES [J].
KUSHNER, HJ ;
YANG, JC .
SIAM JOURNAL ON CONTROL AND OPTIMIZATION, 1993, 31 (04) :1045-1062
[6]  
Maryak JL, 1997, IEEE DECIS CONTR P, P2287, DOI 10.1109/CDC.1997.657115
[7]  
Moulines E., 2011, Advances in Neural Information Processing Systems, V24, P451
[8]   ACCELERATION OF STOCHASTIC-APPROXIMATION BY AVERAGING [J].
POLYAK, BT ;
JUDITSKY, AB .
SIAM JOURNAL ON CONTROL AND OPTIMIZATION, 1992, 30 (04) :838-855
[9]  
Rastogi Pushpendre, 2016, 2016 Annual Conference on Information Science and Systems (CISS), P298, DOI 10.1109/CISS.2016.7460518
[10]   A STOCHASTIC APPROXIMATION METHOD [J].
ROBBINS, H ;
MONRO, S .
ANNALS OF MATHEMATICAL STATISTICS, 1951, 22 (03) :400-407