EXACT SOLUTION FOR ONLINE LEARNING IN MULTILAYER NEURAL NETWORKS

被引：143

作者：

SAAD, D ^{[1
]}

SOLLA, SA ^{[1
]}

机构：

[1] NIELS BOHR INST,CONNECT,DK-2100 COPENHAGEN,DENMARK

来源：

PHYSICAL REVIEW LETTERS | 1995年 / 74卷 / 21期

关键词：

D O I：

10.1103/PhysRevLett.74.4337

中图分类号：

O4 [物理学];

学科分类号：

0702 ;

摘要：

We present an analytic solution to the problem of on-line gradient-descent learning for two-layer neural networks with an arbitrary number of hidden units in both teacher and student networks. © 1995 The American Physical Society.

引用

页码：4337 / 4340

页数：4

共 7 条

[1] BIEHL M, 1994, IN PRESS LUND U REPO
[2] Cybenko G., 1989, Mathematics of Control, Signals, and Systems, V2, P303, DOI 10.1007/BF02551274
[3] INFINITE NUMBER OF ORDER PARAMETERS FOR SPIN-GLASSES
PARISI, G
[J]. PHYSICAL REVIEW LETTERS, 1979, 43 (23) : 1754 - 1756
[4] SAAD D, IN PRESS
[5] LEARNING A RULE IN A MULTILAYER NEURAL-NETWORK
SCHWARZE, H
[J]. JOURNAL OF PHYSICS A-MATHEMATICAL AND GENERAL, 1993, 26 (21): : 5781 - 5794
[6] SEUNG HS, 1993, PHYS REV A, V45, P6056
[7] THE STATISTICAL-MECHANICS OF LEARNING A RULE
WATKIN, TLH
RAU, A
BIEHL, M
[J]. REVIEWS OF MODERN PHYSICS, 1993, 65 (02) : 499 - 556

← 1 →