Stochastic Quasi-Gradient Algorithms with Adaptively Controlled Parameters

Uryasev SP (1986). Stochastic Quasi-Gradient Algorithms with Adaptively Controlled Parameters. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-86-032

[img]
Preview
Text
WP-86-032.pdf

Download (792kB) | Preview

Abstract

The paper deals with choosing stepsize and other parameters in stochastic quasi-gradient methods for solving convex problems of stochastic optimization. The principal idea of methods consists in using random estimates of gradients of the objective function to search for the point of extremum. To control algorithm parameters the iterative adaptive procedures are suggested which are quasi-gradient algorithms with respect to parameters. The convergence is proved and the estimates of the rate of convergence of such algorithms are given. The results of computations for several stochastic optimization problems are considered. The paper is part of the research on numerical techniques for stochastic optimization conducted in the Adaptation and Optimization project of the System and Decision Sciences program.

Item Type: Monograph (IIASA Working Paper)
Research Programs: Adaption and Optimization (ADO)
Depositing User: IIASA Import
Date Deposited: 15 Jan 2016 01:57
Last Modified: 20 Jul 2016 12:50
URI: http://pure.iiasa.ac.at/2827

Actions (login required)

View Item View Item

International Institute for Applied Systems Analysis (IIASA)
Schlossplatz 1, A-2361 Laxenburg, Austria
Phone: (+43 2236) 807 0 Fax:(+43 2236) 71 313