Abstract: | Since the introduction of the Genetic Algorithm model in Holland's seminal work, several other evolution based search strategies have been devised either as a derivative of the originalmodel orindependently. Amongst them, CHC, proposed by L. Eshelman, has been reported as being successfully applied in different domains. The underlying idea of this algorithm is to perform a very aggressive search (by always preserving the best individuals) and, at the same time, offsetting the aggressiveness of the search by using a highly disruptive crossover operator. However, if this leads to a point where the population converges, then a heavy mutation phase is applied, restarting the genetic search afterwards. this document reports on the research carried out to investigate the aability of this search strategy to solve hard real world problems, particularly the Travelling Salesman Problem and the Timetabling, by taking advantage of the aggressive search and cataclysmic mutation to thoroughly search the search space.
|