基于平均化的截尾随机逼近算法  

Averaging-based truncated stochastic approximation algorithm

作　　者：刘仁龙[1] 杨建奎[1] 熊世峰[2] 

机构地区：[1]北京邮电大学理学院,北京100876 [2]中国科学院数学与系统科学研究院,北京100190

出　　处：《中国科学院大学学报（中英文）》2015年第3期289-294,共6页Journal of University of Chinese Academy of Sciences

基　　金：国家自然科学基金(11271355;11101050;11471172)资助

摘　　要：考察带有随机干扰线性系统的随机逼近问题.基于Polyak和Juditsky(SIAM J.Control&Optimization,1992,30:838-855)中的平均化加速算法,提出平均化的截尾算法.证明该算法下随机逼近序列的强相合性和渐近正态性.examined. Control ＆ The almost proved. In this work the stochastic approximation problem of perturbed linear systems was Inspired by the averaging-based accelerated algorithm of Polyak and Juditsky（ SIAM J. Optimization, 1992,30 ： 838-855 ） , we propose an averaging-based truncated algorithm. sure convergence and asymptotic normality of the sequence defined by this algorithm are

关 键 词：渐近正态性 线性系统 强相合性 

分 类 号：O226[理学—运筹学与控制论]