Pareto-optimality solution recommendation using a multi-objective artificial wolf-pack algorithm

Chen, Yi and Wang, Zhonglai and Yang, Erfu and Li, Yun (2016) Pareto-optimality solution recommendation using a multi-objective artificial wolf-pack algorithm. In: The 10th International Conference on Software, Knowledge, Information Management and Application (SKIMA 2016), 2016-12-15 - 2016-12-17, Sichuan Province. (https://doi.org/10.1109/SKIMA.2016.7916207)

[thumbnail of Chen-etal-SKIMA2016-pareto-optimality-solution-recommendation-using-multi-objective-artificial]
Preview
Text. Filename: Chen_etal_SKIMA2016_pareto_optimality_solution_recommendation_using_multi_objective_artificial.pdf
Accepted Author Manuscript

Download (1MB)| Preview

Abstract

In practical applications, multi-objective optimisation is one of the most challenging problems that engineers face. For this, Pareto-optimality is the most widely adopted concept, which is a set of optimal trade-offs between conflicting objectives without committing to a recommendation for decision-making. In this paper, a fast approach to Pareto-optimal solution recommendation is developed. It recommends an optimal ranking for decision-makers using a Pareto reliability index. Further, a mean average precision and a mean standard deviation are utilised to gauge the trend of the evolutionary process. A multi-objective artificial wolf-pack algorithm is thus developed to handle the multi-objective problem using a non-dominated sorting method (MAWNS). This is tested in a case study, where the MAWNS is employed as an optimiser for a widely adopted standard test problem, ZDT6. The results show that the proposed method works valuably for the multi-objective optimisations.