Hybrid Self-organizing Migrating Algorithm Based on Estimation of Distribution

A new hybrid self-organizing migrating algorithm based on estimation of distribution (HSOMA) is proposed to resolve the defect of premature convergence in the self-organizing migrating algorithm (SOMA) and improve the search ability of SOMA. In order to make full use of the statistical information on population and increase the diversity of migration behavior, HSOMA introduces the thought of estimation of distribution algorithm (FDA) into SOMA and reproduces the genes of new individuals by both SOMA and FDA. The proportion of the use of two algorithms is decided by a control parameter. In this way, HSOMA can increase the population diversity and improve the convergence speed. HSOMA is tested on several complex benchmark functions taken from literature and its efficiency is compared with SOMA, the continuous domain Population-Based Incremental Learning algorithm(PBILc) and hybrid migrating behavior based self-organizing migrating algorithm(HBSOMA). On the basis of comparison it is concluded that HSOMA shows better global search ability and convergence accuracy.