Learning in higher order Boltzmann machines using linear response

被引：9

作者：

Leisink, MAR ^{[1
]}

Kappen, HJ ^{[1
]}

机构：

[1] Univ Nijmegen, Dept Biophys, NL-6525 EZ Nijmegen, Netherlands

来源：

NEURAL NETWORKS | 2000年 / 13卷 / 03期

关键词：

higher order Boltzmann machines; linear response; inference; learning;

D O I：

10.1016/S0893-6080(00)00011-3

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

We introduce an efficient method for learning and inference in higher order Boltzmann machines. The method is based on mean field theory with the linear response correction. We compute the correlations using the exact and the approximated method for a fully connected third order network of ten neurons. In addition, we compare the results of the exact and approximate learning algorithm. Finally we use the presented method to solve the shifter problem. We conclude that the linear response approximation gives good results as long as the couplings are not too large. (C) 2000 Elsevier Science Ltd. All rights reserved.

引用

页码：329 / 335

页数：7

共 18 条

[1]

ACKLEY DH, 1985, COGNITIVE SCI, V9, P147

[2] THE HIGH-ORDER BOLTZMANN MACHINE - LEARNED DISTRIBUTION AND TOPOLOGY [J].

ALBIZURI, FX ;

DANJOU, A ;

GRANA, M ;

TORREALDEA, J ;

HERNANDEZ, MC .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 1995, 6 (03) :767-770

[3] INFORMATION GEOMETRY OF BOLTZMANN MACHINES [J].

AMARI, S ;

KURATA, K ;

NAGAOKA, H .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 1992, 3 (02) :260-271

[4]

[Anonymous], 1986, PARALLEL DISTRIBUTED

[5]

BARBER D, 1998, ICANN 98, V1, P93

[6]

BRONSHTEIN IN, 1979, HDB MATH, P430

[7]

Kappen H. J., 1998, P NIPS, P280

[8] Efficient learning in Boltzmann machines using linear response theory [J].

Kappen, HJ ;

Rodriguez, FB .

NEURAL COMPUTATION, 1998, 10 (05) :1137-1156

[9]

KULLBACK S, 1959, INFORMATION THEORY S

[10]

LAWRENCE ND, 1998, P NIPS

← 1 2 →