DocumentCode
769772
Title
An alternative proof for convergence of stochastic approximation algorithms
Author
Kulkarni, S.R. ; Horn, C.S.
Author_Institution
Dept. of Electr. Eng., Princeton Univ., NJ, USA
Volume
41
Issue
3
fYear
1996
fDate
3/1/1996 12:00:00 AM
Firstpage
419
Lastpage
424
Abstract
An alternative proof for convergence of stochastic approximation algorithms is provided. The proof is completely deterministic, very elementary (involving only basic notions of convergence), and direct in that it remains in a discrete setting. An alternative form of the Kushner-Clark condition is introduced and utilized and the results are the first to prove necessity for general gain sequences in a Hilbert space setting
Keywords
Hilbert spaces; approximation theory; convergence of numerical methods; Hilbert space; convergence; deterministic proof; general gain sequences; stochastic approximation algorithms; Adaptive control; Algorithm design and analysis; Approximation algorithms; Convergence; Differential equations; Hilbert space; Stochastic processes; Stochastic resonance; Stochastic systems; System identification;
fLanguage
English
Journal_Title
Automatic Control, IEEE Transactions on
Publisher
ieee
ISSN
0018-9286
Type
jour
DOI
10.1109/9.486642
Filename
486642
Link To Document