Using input parameter influences to support the decisions of feedforward neural networks

被引：41

作者：

Howes, P ^{[1
]}

Crook, N ^{[1
]}

机构：

[1] Oxford Brookes Univ, Sch Comp & Math Sci, Oxford OX3 0BP, England

来源：

NEUROCOMPUTING | 1999年 / 24卷 / 1-3期

关键词：

neural network; parameter influence; rule extraction; explanation; sensitivity analysis;

D O I：

10.1016/S0925-2312(98)00102-7

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Whilst rules extracted from neural networks assist the explanation process, in isolation they do not illustrate the relative importance of each input parameter nor how sensitive the network's output is to these parameters. In this paper we discuss three related measures of input parameter influence which can be used to support explanation facilities for neural networks. An algorithm for generating rules from real-valued networks based on the in influence measures is also presented. (C) 1999 Elsevier Science B.V. All rights reserved.

引用

页码：191 / 206

页数：16

共 34 条

[1] Survey and critique of techniques for extracting rules from trained artificial neural networks [J].

Andrews, R ;

Diederich, J ;

Tickle, AB .

KNOWLEDGE-BASED SYSTEMS, 1995, 8 (06) :373-389

[2] DETERMINING INPUT FEATURES FOR MULTILAYER PERCEPTRONS [J].

BELUE, LM ;

BAUER, KW .

NEUROCOMPUTING, 1995, 7 (02) :111-121

[3] FAST GENETIC SELECTION OF FEATURES FOR NEURAL NETWORK CLASSIFIERS [J].

BRILL, FZ ;

BROWN, DE ;

MARTIN, WN .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 1992, 3 (02) :324-328

[4]

CAWSEY A, 1992, EXPLANATION INTERACT

[5]

Collins E., 1988, IEEE International Conference on Neural Networks (IEEE Cat. No.88CH2632-8), P459, DOI 10.1109/ICNN.1988.23960

[6]

Craven M. W., 1993, P 10 INT C MACH LEAR, P73

[7]

CRAVEN MW, 1996, ADV NEURAL INFORMATI, V8

[8]

CRAVEN MW, P 11 INT C, P37

[9]

CROOK NT, 1992, THESIS OXFORD BROOKE

[10] EXPLANATION AND ARTIFICIAL NEURAL NETWORKS [J].

DIEDERICH, J .

INTERNATIONAL JOURNAL OF MAN-MACHINE STUDIES, 1992, 37 (03) :335-355

← 1 2 3 4 →