MEMORIZATION WITHOUT GENERALIZATION IN A MULTILAYERED NEURAL NETWORK

被引：29

作者：

HANSEL, D

MATO, G

MEUNIER, C

机构：

[1] Centre de Physique Théorique, Ecole Polytechnique, Palaiseau

来源：

EUROPHYSICS LETTERS | 1992年 / 20卷 / 05期

关键词：

GENERAL; THEORETICAL; AND MATHEMATICAL BIOPHYSICS (INC LOGIC OF BIOSYSTEMS; QUANTUM BIOLOGY AND RELEVANT ASPECTS OF THERMODYNAMICS; INFORMATION THEORY; CYBERNETICS AND BIONICS); OTHER TOPICS IN STATISTICAL PHYSICS AND THERMODYNAMICS;

D O I：

10.1209/0295-5075/20/5/015

中图分类号：

O4 [物理学];

学科分类号：

0702 ;

摘要：

The supervised learning of a rule that can be realized by a multilayer network (the teacher) functioning as a parity machine with K = 2 hidden units and nonoverlapping receptive fields is studied. The student network is supposed to have the same architecture as the teacher. Application of statistical mechanics shows that when the number of examples is smaller than a critical value P* the trained network is unable to generalize the rule from the examples. Numerical simulations exhibiting this phenomenon are discussed.

引用

页码：471 / 476

页数：6

共 17 条

[1] STATISTICAL-MECHANICS OF A MULTILAYERED NEURAL NETWORK [J].

BARKAI, E ;

HANSEL, D ;

KANTER, I .

PHYSICAL REVIEW LETTERS, 1990, 65 (18) :2312-2315

[2] BROKEN SYMMETRIES IN MULTILAYERED PERCEPTRONS [J].

BARKAI, E ;

HANSEL, D ;

SOMPOLINSKY, H .

PHYSICAL REVIEW A, 1992, 45 (06) :4146-4161

[3]

BARKAI E, UNPUB

[4]

BAUM E, 1990, IEEE T NEURAL NETWOR

[5] OPTIMAL STORAGE PROPERTIES OF NEURAL NETWORK MODELS [J].

GARDNER, E ;

DERRIDA, B .

JOURNAL OF PHYSICS A-MATHEMATICAL AND GENERAL, 1988, 21 (01) :271-284

[6] THE SPACE OF INTERACTIONS IN NEURAL NETWORK MODELS [J].

GARDNER, E .

JOURNAL OF PHYSICS A-MATHEMATICAL AND GENERAL, 1988, 21 (01) :257-270

[7] 3 UNFINISHED WORKS ON THE OPTIMAL STORAGE CAPACITY OF NETWORKS [J].

GARDNER, E ;

DERRIDA, B .

JOURNAL OF PHYSICS A-MATHEMATICAL AND GENERAL, 1989, 22 (12) :1983-1994

[8]

Gyorgyi G., 1990, NEURAL NETWORKS SPIN, P3, DOI [10.1142/0938, DOI 10.1142/0938]

[9] LEARNING FROM EXAMPLES IN A SINGLE-LAYER NEURAL NETWORK [J].

HANSEL, D ;

SOMPOLINSKY, H .

EUROPHYSICS LETTERS, 1990, 11 (07) :687-692

[10]

MATO G, 1992, IN PRESS J PHYS A

← 1 2 →