Avoidance of traps in stochastic approximation

Borkar, V. S. (2003) Avoidance of traps in stochastic approximation Systems & Control Letters, 50 (1). pp. 1-9. ISSN 0167-6911

Full text not available from this repository.

Official URL: http://linkinghub.elsevier.com/retrieve/pii/S01676...

Related URL: http://dx.doi.org/10.1016/S0167-6911(03)00118-X

Abstract

An estimate for the 'lock-in'probability of stochastic approximation, i.e., the probability of convergence to an attractor given that the algorithm is in its domain of attraction after a sufficiently long time, is derived and is used for proving a.s. avoidance of traps, i.e., convergence only to the stable attractors.

Item Type:Article
Source:Copyright of this article belongs to Elsevier Science.
Keywords:Stochastic Approximation; O. D. E. Limit; Lock-in Probability; Avoidance of Traps; Kushner-clark Lemma
ID Code:5280
Deposited On:18 Oct 2010 07:54
Last Modified:20 May 2011 08:58

Repository Staff Only: item control page