Adaptive Systems

Non-Markov algorithms for statistical optimization with continuous time. II

O. Yu. Kulchitskii

Leningrad

Abstract: A procedure is proposed for estimating the convergence rates of statistical optimization algorithms with continuous time of non-Markov type. Estimates of convergence rate asymptotic as $t\to\infty$ of such Robbins — Monro, Keefer — Wolfowitz, and random search algorithms.

UDC: 62-505:518.5


Received: 12.05.1977

 English version: Automation and Remote Control, 1978, 39:6, 823–832

Bibliographic databases:

•