Evolutionary Programming using a mixed strategy with incomplete information

Abstract

Evolutionary Programming (EP) has been modified in various ways. In particular, modifications of the mutation operator have been proved to be capable of significantly improving the performance of EP. However, while each of proposed mutation operators (e.g. Gaussian mutation and Cauchy mutation) may be suitable for solving certain types of problem, none of them are suitable for all problems. Mixed strategies have therefore been proposed in order to combine the advantages of different operators. The design of a mixed strategy is currently based on the premise that complete and perfect information is held for each mutation operator in the mixed strategy such that the payoff functions to each pure strategy are common knowledge. This paper presents a modified mixed strategy (IMEP) involving a process with incomplete information. Experimental results show that IMEP outperforms pure strategy algorithms in spite of the lack of information. The experiments also show that the results are similar to those generated by the original algorithm, which was complete information.