PROJECT TITLE :

Generalization of Pareto-Optimality for Many-Objective Evolutionary Optimization

ABSTRACT:

The vast majority of multiobjective evolutionary algorithms presented up to now are Pareto-primarily based. Sometimes, these algorithms perform well for issues with few (2 or 3) objectives. But, thanks to the poor discriminability of Pareto-optimality in several-objective spaces (usually four or additional objectives), their effectiveness deteriorates progressively as the problem dimension will increase. This paper generalizes Pareto-optimality each symmetrically and asymmetrically by expanding the dominance area of solutions to enhance the scalability of existing Pareto-based mostly algorithms. The generalized Pareto-optimality (GPO) criteria are comparatively studied in terms of the distribution of ranks, the ranking landscape, and therefore the convergence of the evolutionary process over several benchmark issues. The results indicate that algorithms equipped with a generalized optimality criterion will acquire the pliability of adjusting their choice pressure at intervals bound ranges, and achieve a richer variety of ranks to realize faster and higher convergence on some subsets of the Pareto optima. To make amends for the doable diversity loss induced by the generalization, a distributed evolution framework with adaptive parameter setting is also proposed and briefly discussed. Empirical results indicate that this strategy is sort of promising in diversity preservation for algorithms associated with the GPO.