The Annals of Statistics

Nonlinear Stochastic Approximation Procedures for $L_p$ Loss Functions

Zhiliang Ying

Full-text: Open access

Abstract

The classical stochastic approximation problem can be regarded as choosing design points so that the responses are close to some target level in the expected squared distance. Motivated by different loss criteria, a family of stochastic approximation algorithms is proposed. This family has the same simplicity as the classical Robbins-Monro procedure does and contains the latter as a special case. Using appropriate representations and martingale limit theorems, we establish asymptotic properties for this family. Using the semiparametric formulation, lower bounds are obtained for estimating the desired parameters under any adaptive design, showing that the proposed algorithms with appropriate scaling are asymptotically efficient.

Article information

Source
Ann. Statist., Volume 18, Number 4 (1990), 1817-1828.

Dates
First available in Project Euclid: 12 April 2007

Permanent link to this document
https://projecteuclid.org/euclid.aos/1176347880

Digital Object Identifier
doi:10.1214/aos/1176347880

Mathematical Reviews number (MathSciNet)
MR1074437

Zentralblatt MATH identifier
0716.62080

JSTOR
links.jstor.org

Subjects
Primary: 62L20: Stochastic approximation
Secondary: 62L05: Sequential design 60F17: Functional limit theorems; invariance principles

Keywords
Stochastic approximation Robbins-Monro procedure sequential design $L_p$ loss information bound

Citation

Ying, Zhiliang. Nonlinear Stochastic Approximation Procedures for $L_p$ Loss Functions. Ann. Statist. 18 (1990), no. 4, 1817--1828. doi:10.1214/aos/1176347880. https://projecteuclid.org/euclid.aos/1176347880


Export citation