AL-Madi, Nagham Azmi and Khader, Ahamad Tajudin (2007) A Social Based Model For Genetic Algorithms. In: Proceedings Of The Third International Conference On Information Technology (ICIT), 9-11 May 2007, Zaytoonah University, Amman, Jordan.
Genetic algorithms (GAs), as a general search model, have proved its success in several applications, however, recently, several researchers have argued that they have slow convergence; this slowness is due to the randomness in all their operations. Therefore, recent researches have employed structured populations, in order to eliminate randomness, such as island models, cellular model, multinational evolutionary algorithms, etc. In this proposal, a social based GA is introduced; this model is trying to mimic the actual social behavior and the actual death and birth process. We will restrict the recombination for males to the only permitted females; we also divide the population into nearly separated subgroups (similar to the island model). Our motivation to such an approach is that we expect the nature to be more robust and optimal; hence the objectives of this work are to study the effects of these social rules and customs on the standard GA, and to investigate its effects on the speed of convergence of GA. The results will be analyzed according to parameters that depend on the social behavior and the natural birth and death models.
Actions (login required)