Adaptive Systems

On the rate of convergence and choice of parameters for an automaton algorithm

A. V. Nazin

Moscow

Abstract: In the problem of adaptive choice of options the mean losses have to be minimized. An automaton algorithm is proposed for its solution which is an extension of the well-known algorithm of the method of stochastic approximation. Mean square convergence conditions are derived; the rate of convergence is estimated; and optimal and guaranteeing algorithm parameters are determined.

UDC: 65.01


Received: 10.03.1981

 English version: Automation and Remote Control, 1982, 43:7, 896–905

Bibliographic databases:

• 
•