Neural classifier construction using regularization, pruning and test error estimation

被引:18
作者
Hintz-Madsen, M [1 ]
Hansen, LK [1 ]
Larsen, J [1 ]
Pedersen, MW [1 ]
Larsen, M [1 ]
机构
[1] Tech Univ Denmark, Dept Math Modelling, CONNECT, DK-2800 Lyngby, Denmark
关键词
neural classifiers; architecture optimization; regularization; generalization estimation;
D O I
10.1016/S0893-6080(98)00093-8
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this paper we propose a method for construction of feed-forward neural classifiers based on regularization and adaptive architectures. Using a penalized maximum likelihood scheme, we derive a modified form of the entropic error measure and an algebraic estimate of the test error. In conjunction with optimal brain damage pruning, a test error estimate is used to select the network architecture. The scheme is evaluated on four classification problems. (C) 1998 Elsevier Science Ltd. All rights reserved.
引用
收藏
页码:1659 / 1670
页数:12
相关论文
共 21 条
[1]  
ANDERSEN LN, 1997, P IEEE WORKSH NEUR N, V7, P24
[2]  
Bishop C. M., 1995, NEURAL NETWORKS PATT
[3]  
BRIDLE J, 1990, NEUROCOMPUTING ALGOR, V6, P227
[4]  
DENKER J, 1987, COMPLEX SYSTEMS, V2, P877
[5]  
Gorodkin J, 1993, Int J Neural Syst, V4, P159, DOI 10.1142/S0129065793000146
[6]   FOURIER PREPROCESSING FOR HAND PRINT CHARACTER RECOGNITION [J].
GRANLUND, GH .
IEEE TRANSACTIONS ON COMPUTERS, 1972, C 21 (02) :195-+
[7]  
Hart P.E., 1973, Pattern recognition and scene analysis
[8]  
Hintz-madsen M., 1995, P 1995 IEEE WORKSH N, P484
[9]  
LARSEN J, 1996, P IEEE WORKSH NEUR N, V6, P62
[10]  
LARSEN J, 1993, THESIS TU DENMARK