Fuzzy Number Neural Networks
This paper presents a practical algorithm for training neural networks with fuzzy number weights, inputs, and outputs. Typically, fuzzy number neural networks are difficult to train because of the many α-cut constraints implied by the fuzzy weights. A transformation is used to eliminate these constraints and allow use of standard unconstrained optimization methods. The algorithm is demonstrated on a three-layer network.
J. P. Dunyak and D. C. Wunsch, "Fuzzy Number Neural Networks," Fuzzy Sets and Systems, vol. 108, no. 1, pp. 49-58, Elsevier, Jan 1999.
The definitive version is available at http://dx.doi.org/10.1016/S0165-0114(97)00339-4
Electrical and Computer Engineering
International Standard Serial Number (ISSN)
Article - Journal
© 1999 Elsevier, All rights reserved.