Techniques for Highly Multiobjective Optimisation: Some Nondominated Points are Better than Others - Computer Science > Neural and Evolutionary ComputingReport as inadecuate




Techniques for Highly Multiobjective Optimisation: Some Nondominated Points are Better than Others - Computer Science > Neural and Evolutionary Computing - Download this document for free, or read online. Document in PDF available to download.

Abstract: The research area of evolutionary multiobjective optimization EMO isreaching better understandings of the properties and capabilities of EMOalgorithms, and accumulating much evidence of their worth in practicalscenarios. An urgent emerging issue is that the favoured EMO algorithms scalepoorly when problems have many e.g. five or more objectives. One of the chiefreasons for this is believed to be that, in many-objective EMO search,populations are likely to be largely composed of nondominated solutions. Inturn, this means that the commonly-used algorithms cannot distinguish betweenthese for selective purposes. However, there are methods that can be usedvalidly to rank points in a nondominated set, and may therefore usefullyunderpin selection in EMO search. Here we discuss and compare several suchmethods. Our main finding is that simple variants of the often-overlookedAverage Ranking strategy usually outperform other methods tested, coveringproblems with 5-20 objectives and differing amounts of inter-objectivecorrelation.



Author: David Corne, Joshua Knowles

Source: https://arxiv.org/







Related documents