Traditional random search methods incorporate symmetric density functions for generating perturbations about the operating point. This philosophy has been retained in the development of multi-agent stochastic search techniques including methods in evolutionary computation. The present work introduces asymmetric mutations for stochastic search. The asymmetric mutations are generated via a probabilistic switching mechanism that biases the search based on self-adaptive strategy parameters. The dynamics of the strategy parameters are explored and then investigated in light of using asymmetric perturbations.
展开▼
