Local greedy approximation for nonlinear regression and neural network training

• 成果类型: Article
• 署名作者: Jones, LK
• 署名单位: University of Massachusetts System; University of Massachusetts Lowell
• 刊物名称: ANNALS OF STATISTICS
• ISSN/ISSBN: 0090-5364
• DOI: 10.1214/aos/1015957398
• 发表日期: 2000
• 页码: 1379-1389
• 关键词: projection pursuit regression; pattern-classification; CONVERGENCE; algorithms

摘要

A criterion for local estimation and approximation in nonlinear regression and neural network training is introduced and motivated. Nth-order greedy approximation for the regression (or target) function based on the criterion is shown to converge at rate O(1/N-1/2) in the nonsampling case.