Application of Fuzzy Automata to Fuzzy Signal Processing

For better fuzzy signal processing, this paper presents a new processing tool-fuzzy automata, discusses mainly how the fuzzy automata infer the fuzzy logic, and how the fuzzy automata can be obtained by using the recurrent neural networks to fuzzy logic processing. At first, the fuzzy logical rules are introduced, and the definition of fuzzy finite state automata is also given. In addition, the fuzzy knowledge equivalence representations among neural networks, the fuzzy systems and models of automata are discussed. Once the networks have been trained, we will develop a method to extract a representation of the fuzzy finite state automaton (FFA) encoded in the recurrent neural networks for recognizing the training rules. Moreover, an example and application for fuzzy signal processing are given.