Representation and Identification of Finite State Automata by Recurrent Neural Networks

This paper presents a new architecture of neural networks for representing deterministic finite state automata. The proposed model is capable of strictly representing FSA with the network size being smaller than the existing models proposed so far. We also discuss an identification method of FSA from a given set of input and output data by training the proposed neural networks. We apply the genetic algorithm to determine the parameters of the neural network such that its input and output relation becomes equal to those of a target FSA.