PHONEME-BASED SPEECH RECOGNITION VIA FUZZY NEURAL NETWORKS MODELING AND LEARNING

Fuzzy neural networks (FNN) have several features that make them well suited to a wide range of knowledge engineering applications. These st rengths include fast and accurate learning, good generalisation capabi lities, excellent explanation facilities in the form of semantically m eaningful fuzzy rules, and the ability to accommodate both data and ex isting expert knowledge about the problem under consideration. The pap er presents one particular architecture called FuNN and discusses two alternative ways to optimise its structure, namely a genetic algorithm and a method of learning-with-forgetting. The optimised structure has much less connections and can easily be interpreted in terms of fuzzy rules. Such a structure can be effectively used for on-line adaptatio n which is demonstrated on a phoneme-based speech recognition problem. (C) 1998 Elsevier Science Inc. All rights reserved.