LEARNING BIAS IN NEURAL NETWORKS AND AN APPROACH TO CONTROLLING ITS EFFECTS IN MONOTONIC CLASSIFICATION

被引：8

作者：

ARCHER, NP ^{[1
]}

WANG, SH ^{[1
]}

机构：

[1] UNIV NEW BRUNSWICK,FAC BUSINESS,ST JOHN E2L 4L5,NB,CANADA

来源：

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE | 1993年 / 15卷 / 09期

关键词：

BACKPROPAGATION; LEARNING BIAS; MONOTONICALLY SEPARABLE; MONOTONIC BOUNDARY; MONOTONICITY; NEURAL NETWORK;

D O I：

10.1109/34.232084

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

As a learning machine, a neural network using the back-propagation training algorithm is subject to learning bias. This results in unpredictability of boundary generation behavior in pattern recognition applications, especially in the case of small training sample size. This research sugests that in a large class of pattern recognition problems, such as managerial and other problems possessing monotonicity properties, the effect of learning bias can be controlled by using multiarchitecture monotonic function neural networks.

引用

页码：962 / 966

页数：5

共 20 条

[11] KOHONEN T, 1988, P 1988 IEEE INT C NE, pI61
[12] RAMANUJAM V, 1986, ACAD MANAGE J, V29, P347, DOI 10.2307/256192
[13] Rumelhart DE, 1986, ENCY DATABASE SYST, P45
[14] SOULIE FF, 1987, 1ST P IEEE INT C NEU
[15] Specht D F, 1990, IEEE Trans Neural Netw, V1, P111, DOI 10.1109/72.80210
[16] PROBABILISTIC NEURAL NETWORKS
SPECHT, DF
[J]. NEURAL NETWORKS, 1990, 3 (01) : 109 - 118
[17] Utgoff P., 1986, MACHINE LEARNING ART, VII, P107
[18] Utgoff P.E., 1986, MACHINE LEARNING IND
[19] WIELAND A, 1987, 1ST P IEEE INT C NEU
[20] [No title captured]

← 1 2 →