多维情形下Kesten随机逼近算法推广的几乎必然收敛与渐近正态性
统计理论
2011-05-27 v1 统计理论
摘要
本文证明了多维情形下Kesten随机逼近算法推广的几乎必然收敛与渐近正态性。在该推广中,若估计的两次相继增量的标量积为正或负,则步长相应增加或减小。此规则旨在当初始条件导致算法在起始迭代中表现为“确定性方式”时,加速其进入“随机行为”。
关键词
引用
@article{arxiv.1105.5231,
title = {Almost sure convergence and asymptotical normality of a generalization of Kesten's stochastic approximation algorithm for multidimensional case},
author = {Pedro Cruz},
journal= {arXiv preprint arXiv:1105.5231},
year = {2011}
}
备注
25 pages, 1 figure