Knowledge Extraction From Neural Networks Using the All-Permutations Fuzzy Rule Base: The LED Display Recognition Problem

Author

Kolman, E. ; Margaliot, M.

Author_Institution

Sch. of Electr. Eng.-Syst., Tel Aviv Univ.

Volume

18

Issue

3

fYear

2007

fDate

5/1/2007 12:00:00 AM

Firstpage

925

Lastpage

931

Abstract

A major drawback of artificial neural networks (ANNs) is their black-box character. Even when the trained network performs adequately, it is very difficult to understand its operation. In this letter, we use the mathematical equivalence between ANNs and a specific fuzzy rule base to extract the knowledge embedded in the network. We demonstrate this using a benchmark problem: the recognition of digits produced by a light emitting diode (LED) device. The method provides a symbolic and comprehensible description of the knowledge learned by the network during its training

Keywords

LED displays; fuzzy set theory; knowledge acquisition; neural nets; LED display recognition problem; all-permutations fuzzy rule base; artificial neural networks; black-box character; knowledge extraction; light emitting diode device; Artificial neural networks; Data mining; Displays; Fuzzy neural networks; Fuzzy systems; Hybrid intelligent systems; Knowledge based systems; Knowledge representation; Light emitting diodes; Neural networks; Feedforward neural networks; hybrid intelligent systems; knowledge extraction; neurofuzzy systems; rule extraction; rule generation; Algorithms; Artificial Intelligence; Computer Simulation; Decision Support Techniques; Information Storage and Retrieval; Models, Theoretical; Neural Networks (Computer); Pattern Recognition, Automated;

fLanguage

English

Journal_Title

Neural Networks, IEEE Transactions on

Publisher

ieee

ISSN

1045-9227

Type

jour

DOI

10.1109/TNN.2007.891686

Filename

4182389