Population-less Genetic Algorithm? Investigation of Non-dominated Tournament Genetic Algorithm (NTGA2) for multi-objective optimization
Michał Antkiewicz, Pawel Myszkowski
DOI: http://dx.doi.org/10.15439/2023F3491
Citation: Communication Papers of the 18th Conference on Computer Science and Intelligence Systems, M. Ganzha, L. Maciaszek, M. Paprzycki, D. Ślęzak (eds). ACSIS, Vol. 37, pages 21–28 (2023)
Abstract. This paper investigates the Non-dominated Tournament Genetic Algorithm (NTGA2) to examine how selection methods (and population) interact in solving multi-objective optimization problems with constraints. As NTGA2 uses tournament and GAP selections that link the current population and population' archive, the experiments' results show that the population role is significantly reduced in some cases. The study considers two benchmark problems: Multi-Skill Resource Constrained Project Scheduling Problem (MS-RCPSP) and Travelling Thief Problem. Moreover, the paper's experimental study consists of new instances for multi-objective MS-RCPSP to show some interesting results that, in some cases, the proposed Genetic Algorithm does not need population in the evolution process.