GENERALIZATION ABILITY OF PERCEPTRONS WITH CONTINUOUS OUTPUTS

被引:21
作者
BOS, S
KINZEL, W
OPPER, M
机构
[1] Institut für Theoretische Physik, Justus-Liebig Universität, D-6300 Giessen
来源
PHYSICAL REVIEW E | 1993年 / 47卷 / 02期
关键词
D O I
10.1103/PhysRevE.47.1384
中图分类号
O35 [流体力学]; O53 [等离子体物理学];
学科分类号
070204 ; 080103 ; 080704 ;
摘要
In this paper we examine the influence of different input-output relations on the generalization ability of a single-layer perceptron. The input-output relations can be linear, binary, or sigmoid. With this choice we take into account most of the cases which are of present interest. The generalization problem will be realizable or unrealizable if the input-output relations for teacher and student are identical or not. We show that sometimes it can have a positive effect on the generalization ability, if one learns with errors.
引用
收藏
页码:1384 / 1391
页数:8
相关论文
共 19 条
[11]  
OPPER M, 1991, 4TH P ANN ACM WORKSH
[12]  
Rosenblatt F, 1962, PRINCIPLES NEURODYNA
[13]  
Rumelhart D.E., 1986, PARALLEL DISTRIBUTED, DOI 10.7551/mitpress/5236.001.0001
[14]  
Rumelhart DE, 1986, ENCY DATABASE SYST, P45
[15]   STATISTICAL-MECHANICS OF LEARNING FROM EXAMPLES [J].
SEUNG, HS ;
SOMPOLINSKY, H ;
TISHBY, N .
PHYSICAL REVIEW A, 1992, 45 (08) :6056-6091
[16]  
SEUNG HS, 1992, 5TH P ANN ACM WORKSH
[17]  
SEUNG HS, COMMUNICATION
[18]   LINEAR AND NONLINEAR EXTENSION OF THE PSEUDO-INVERSE SOLUTION FOR LEARNING BOOLEAN FUNCTIONS [J].
VALLET, F ;
CAILTON, JG ;
REFREGIER, P .
EUROPHYSICS LETTERS, 1989, 9 (04) :315-320
[19]   SELECTING EXAMPLES FOR PERCEPTRONS [J].
WATKIN, TLH ;
RAU, A .
JOURNAL OF PHYSICS A-MATHEMATICAL AND GENERAL, 1992, 25 (01) :113-121