An Improved Genetic Algorithm for Finite-State-Machine Evolution

The Finite-State-Machine (FSM) evolution using Genetic Algorithm (GA) is a typical theme in Evolvable Hardware (EHW) research. In this paper, an improved GA is described for FSM evolution. According to the characteristic that output is relevant to state transition in FSM, this algorithm divides the chromosome representing FSM into state transition gene and output gene for stage-by-stage evolution. Experimental results indicate that this improved GA greatly decreases the complexity of FSM evolution and gets better performance.