Revision as of 16:11, 3 May 2008 edit NS11 (talk \| contribs) 3 edits made it more readable, in Robbins-Monro algorithm fix $(.....)/c_n$ to $(.....)/2c_n$, change family of optimization algs to family of algs ← Previous edit		Revision as of 19:10, 4 May 2008 edit undo Spacepotato (talk \| contribs) Autopatrolled, Extended confirmed users, Pending changes reviewers, Rollbackers 27,611 edits They're stochastic optimization algorithms. Next edit →
Line 1: '''Stochastic approximation''' methods are a family of iterative ~~algorithms~~[[stochastic optimization]] [[algorithm]]s that attempt to find zeroes or extrema of functions which cannot be computed directly, but only estimated via noisy observations. The first, and prototypical, algorithms of this kind were the '''Robbins-Monro''' and '''Kiefer-Wolfowitz''' algorithms. == Robbins-Monro algorithm ==

Stochastic approximation: Difference between revisions