DIMENSION-INDEPENDENT BOUNDS ON THE DEGREE OF APPROXIMATION BY NEURAL NETWORKS

作     者：MHASKAR, HN MICCHELLI, CA 

作者机构：Department of Mathematics and Computer Science California State University Los Angeles 90032 USA|c| 

出 版 物：《IBM JOURNAL OF RESEARCH AND DEVELOPMENT》 (国际商用机器公司研究与开发杂志)

年 卷 期：1994年第38卷第3期

页      面：277-284页

核心收录：

学科分类：08[工学] 0835[工学-软件工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

主　　题：Neural networks 

摘      要：Let phi be a univariate 2pi-periodic function. Suppose that s greater-than-or-equal-to 1 and f is a 2pi-periodic function of s real variables. We study sufficient conditions in order that a neural network having a single hidden layer consisting of n neurons, each with an activation function phi, can be constructed so as to give a mean square approximation to f within a given accuracy epsilon(n), independent of the number of variables. We also discuss the case in which the activation function phi is not 2pi-periodic.