By Topic

Noise conditions for prespecified convergence rates of stochastic approximation algorithms

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

3 Author(s)
E. K. P. Chong ; Sch. of Electr. Eng., Purdue Univ., West Lafayette, IN, USA ; I-Jeng Wang ; S. R. Kulkarni

We develop deterministic necessary and sufficient conditions on individual noise sequences of a stochastic approximation algorithm for the error of the iterates to converge at a given rate. Specifically, suppose {ρn} is a given positive sequence converging monotonically to zero. Consider a stochastic approximation algorithm x n+1=xn-an(Anxn-b n)+anen, where {xn} is the iterate sequence, {an} is the step size sequence, {en } is the noise sequence, and x* is the desired zero of the function f(x)=Ax-b. Then, under appropriate assumptions, we show that x n-x*=o(ρn) if and only if the sequence {en} satisfies one of five equivalent conditions. These conditions are based on well-known formulas for noise sequences: Kushner and Clark's (1978) condition, Chen's (see Proc. IFAC World Congr., p.375-80, 1996) condition, Kulkarni and Horn's (see IEEE Trails Automat. Contr., vol.41, p.419-24, 1996) condition, a decomposition condition, and a weighted averaging condition. Our necessary and sufficient condition on {en} to achieve a convergence rate of {ρn} is basically that the sequence {enn} satisfies any one of the above five well-known conditions. We provide examples to illustrate our result. In particular, we easily recover the familiar result that if an=a/n and {en} is a martingale difference process with bounded variance, then xn-x*=o(n-1/2(log(n))β ) for any β>1/2

Published in:

IEEE Transactions on Information Theory  (Volume:45 ,  Issue: 2 )