ON BATHER STOCHASTIC-APPROXIMATION ALGORITHM

Stochastic approximation procedures provide a useful technique for detecting the root of an unknown regression function. Based on the idea of averaging Bather (1989) proposed a new stochastic approximation algorithm at the Fourth Prague Symposium on Asymptotic Statistics. For this algorithm some results will be presented on the rate of convergence as well as on the behaviour for small to moderate sample sizes.