Supplemental materials for the paper "Interactive one-max problem allows to compare the performance of interactive and human-based genetic algorithms"

Abstract

Human-based genetic algorithm (HBGA) uses both human evaluation and human innovation to optimize a population of solutions (Kosorukoff, 2001). The novel contribution of HBGA is an introduction of human-based innovation operators. However, there was no attempt to measure the effect of those human-based innovation operators on the overall performance of a GA quantitatively, in particular, by comparing the performance of HBGA and interactive genetic algorithm (IGA) that do not use human innovation. This paper shows that the mentioned effect is measurable and further focuses on quantitative comparison of the efficiency of these two classes of algorithms. In order to achieve this purpose, the paper proposes an interactive analog of the one-max problem, suggests some human-based innovation operators appropriate for this problem, and compares convergence results of an HBGA and an IGA for the same problem.


	Abstract Human-based genetic algorithm (HBGA) uses both human evaluation and human innovation to optimize a population of solutions (Kosorukoff, 2001). The novel contribution of HBGA is an introduction of human-based innovation operators. However, there was no attempt to measure the effect of those human-based innovation operators on the overall performance of a GA quantitatively, in particular, by comparing the performance of HBGA and interactive genetic algorithm (IGA) that do not use human innovation. This paper shows that the mentioned effect is measurable and further focuses on quantitative comparison of the efficiency of these two classes of algorithms. In order to achieve this purpose, the paper proposes an interactive analog of the one-max problem, suggests some human-based innovation operators appropriate for this problem, and compares convergence results of an HBGA and an IGA for the same problem.

	Presentation and Full Paper Full paper Powerpoint presentation Brightness (M1) of the brightest individual in the population for: Generational type of IGA and HBGA (Figure 2) Steady state type of IGA and HBGA (Figure 3) Euclidean metric (M2) of the brightest individual in the population for: Generational type of IGA and HBGA (Figure 4) Steady state type of IGA and HBGA (Figure 5) Minimum Component Metric (MS) of the brightest individual in the population for: Generational type of IGA and HBGA (Figure 6) Steady state type of IGA and HBGA (Figure 7)

	BibTex Cheng, C.D., Kosorukoff, A. (2004). Interactive one-max problem allows to compare the performance of interactive and human-based genetic algorithms, Genetic and Evolutionary Computation – GECCO 2004: Genetic and Evolutionary Computation Conference, Seattle, WA, USA, June 26-30, 2004. Proceedings, Part I , pp. 983-993. @INPROCEEDINGS{CDChengGECCO04, AUTHOR = {Chihyung Derrick Cheng and Alexander Kosorukoff}, TITLE = {Interactive one-max problem allows to compare the performance of interactive and human-based genetic algorithms}, BOOKTITLE = {Genetic and Evolutionary Computation – GECCO 2004: Genetic and Evolutionary Computation Conference, Seattle, WA, USA, June 26-30, 2004. Proceedings, Part I }, EDITOR = {Kalyanmoy Deb, et al}, PUBLISHER = {Springer Verlag (LNCS 3102)}, PLACE = {Berlin}, VOLUME = {1}, PAGES = {983 - 993}, YEAR = {2004}, } \| Go back to content \|