Many objective optimization algorithm based on mixed species particle flocking.

Autor: Anand, H. S., Saritha, R., Vinod Chandra, S. S.
Zdroj: Journal of Ambient Intelligence & Humanized Computing; Oct2023, Vol. 14 Issue 10, p13251-13267, 17p
Abstrakt: A methodology inspired by the combined flocking nature of mixed species of birds is proposed in this paper. Mixed species flock with the intention to increase feeding efficiency or to defend predator attacks. Specific particle orientation in the entire population is controlled by particle's previous best location, position of the best species member and a globally best particle position. Non dominated solution set of the multiple objectives targeted by each species is acquired by tracking the trajectory of the globally best particle during current course of action. The proposed method provides adaptive population strategy by giving robust performance on divergent decision and objective space. Particle depletion and augmentation procedure increases the productivity of population to direct the solution set move closer to true pareto front with maximum spread and diversity. Visualisation of the non dominating set is provided using a solution ranking approach supported by a heap tree. It also acts as an ordered repository of partial solutions along with a quality ranking parameter. Performance of the method is evaluated over a standard set of multi objective test problems from the latest many objective function(MaF) test suit. Quantitative and qualitative assessment of the proposed model with state of the art many objective optimization algorithms in the field of natural computing depicts its endurance on scalable problems of different dynamics without compromising the computational complexity, but with finer storage alternative. [ABSTRACT FROM AUTHOR]
Databáze: Complementary Index