Logo image
Enhanced convergence properties of populations through aging, immunity, and instinct
Conference proceeding   Open access   Peer reviewed

Enhanced convergence properties of populations through aging, immunity, and instinct

Hakki Erhan Sevil, Erkin Gezgin and Serhan Ozdemir
pp.231-238
TAINN 2005, Turkish Symposium on Artificial Intelligence and Neural Networks (Izmir, Turkey, 07/16/2005–07/17/2005)
2005

Metrics

3 File views/ downloads
73 Record Views

Abstract

This study is based on an earlier work [ 1] called Responsive Perturbation Algorithm (RPA), in which a population is created by perturbing a winner or best performing individual in a neighborhood whose boundaries are constantly defined over periods of cycles. One advantage of such a scheme is that all the offsprings are descendents of the best-performing parent, and not the hybrids of the best and the second-best parents. This work sets a population of homogenous individuals, and a population of heterogeneous individuals to compete. The more advanced work is now called, with the supplements, trait-based heterogeneous populations plus (TbHP+). Credit assignment and bias concepts in the form of immunity and instinct have been added to provide the populations with a more efficient guidance. Simulations show that traits are essential to prompt an early and a certain convergence.
pdf
Enhanced convergence properties of populations...689.91 kBDownloadView
PresentationConference paper pdf Open Access

Related links

Details

Logo image