Automata Construct with Genetic Algorithm

A finite state machine (FSM) is a behavioural model of digital hardware. The "revolutionary programming" algorithm was created in 60’s by Fogel and it was the first attempt to construct a FSM by evolutionary principle, but only with usage a mutation operator. Lucas and Reynolds tried to construct FSMs without output function using genetic algorithm (for purpose of lexical analysers). This contribution describes experiments with a standard genetic algorithms (SGA) on more general FSMs (Moore and Mealy). It compares simple GA with those, where the mutation affected by the probability. The minimisation is built-in genetic algorithm. The algorithm is tested on several examples.