Almost sure convergence of stochastic gradient processes with matrix step sizes

被引:6
|
作者
Monnez, JM [1 ]
机构
[1] Univ Henri Poincare, IECN Lab Math, F-54506 Vandoeuvre Les Nancy, France
关键词
stochastic approximation; stochastic gradient;
D O I
10.1016/j.spl.2005.09.014
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
We consider a stochastic gradient process, which is a special case of stochastic approximation process, where the positive real step size a(n) is replaced by a random matrix A(n): Xn+1 = X-n - A(n)del g(X-n) - A(n)V(n). We give two theorems of almost sure convergence in the case where the equation del g = 0 has a set of solutions. (C) 2005 Elsevier B.V. All rights reserved.
引用
收藏
页码:531 / 536
页数:6
相关论文
共 50 条