Fusion of neural networks with fuzzy logic and genetic algorithm

Article type: Research Article

Authors: Cho, Sung-Bae

Affiliations: Department of Computer Science, Yonsei University, 134 Shinchon-dong, Sudaemoon-ku, Seoul 120-749, South Korea Tel.: +82 2 2123 2720; Fax: +82 2 365 2579; E-mail: [email protected]

Abstract: Combining multiple estimators has appeared as one of hot research topics in several areas including artificial neural networks. This paper presents three methods to work out the problem based on so-called softcomputing techniques, toward a unified framework of hybrid softcomputing techniques. The first method based on fuzzy logic nonlinearly combines objective evidences, in the form of network outputs, with subjective evaluation of the reliability of the individual neural networks. The second method based on genetic algorithm gives us an effective vehicle to determine the optimal weight parameters that are multiplied by the network outputs. Finally, we have proposed a hybrid synergistic method of fuzzy logic and genetic algorithm to optimally combine neural networks. The experimental results with the recognition problem of totally unconstrained handwritten digits show that the performance could be improved significantly with the proposed softcomputing techniques.

DOI: 10.3233/ICA-2002-9405

Journal: Integrated Computer-Aided Engineering, vol. 9, no. 4, pp. 363-372, 2002