Title :
Analysis of Some Mating and Collaboration Strategies in Evolutionary Algorithms
Author :
Gog, Anca ; Chira, Camelia ; Dumitrescu, D. ; Zaharie, Daniela
Author_Institution :
Dept. of Comput. Sci., Babes Bolyai Univ., Cluj-Napoca, Romania
Abstract :
The selection of mates in an evolutionary algorithm can significantly influence the exploration and the exploitation abilities of the search process. Currently there are several strategies to guide the mate selection or to restrict the mating pool. The aim of this paper is to analyze the behavior of a fitness guided mate selection strategy and of a collaboration strategy between population elements which use different mating rules. The behavior is analyzed empirically for mate selection in differential evolution algorithms by considering that the population is divided in two subpopulations characterized by different mating rules. The dynamics of these subpopulations sizes is also theoretically analyzed.
Keywords :
evolutionary computation; search problems; behavior analysis; collaboration strategy; differential evolution algorithm; evolutionary algorithm; fitness guided mate selection strategy; population element; search process; Algorithm design and analysis; Computer science; Context modeling; Design optimization; Evolutionary computation; International collaboration; Multiagent systems; Scientific computing; Terminology; Topology; differential evolution; evolutionary algorithms; exploration and exploitation; mating rules; subpopulation;
Conference_Titel :
Symbolic and Numeric Algorithms for Scientific Computing, 2008. SYNASC '08. 10th International Symposium on
Conference_Location :
Timisoara
Print_ISBN :
978-0-7695-3523-4
DOI :
10.1109/SYNASC.2008.87
