Theoretically Investigating Optimal µ-Distributions for the Hypervolume Indicator: First Results for Three Objectives

Several indicator-based evolutionary multiobjective optimization algorithms have been proposed in the literature. The notion of optimal µ-distributions formalizes the optimization goal of such algorithms: find a set of µ solutions that maximizes the underlying indicator among all sets with µ solutions. In particular for the often used hypervolume indicator, optimal µ-distributions have been theoretically analyzed recently. All those results, however, cope with bi-objective problems only. It is the main goal of this paper to extend some of the results to the 3-objective case. This generalization is shown to be not straight-forward as a solution's hypervolume contribution has not a simple geometric shape anymore in opposition to the bi-objective case where it is always rectangular. In addition, we investigate the influence of the reference point on optimal µ-distributions and prove that also in the 3-objective case situations exist for which the Pareto front's extreme points cannot be guaranteed in optimal µ-distributions.

[1]  Anne Auger,et al.  Theory of the hypervolume indicator: optimal μ-distributions and the choice of the reference point , 2009, FOGA '09.

[2]  Nicola Beume,et al.  On the Complexity of Computing the Hypervolume Indicator , 2009, IEEE Transactions on Evolutionary Computation.

[3]  Nicola Beume,et al.  SMS-EMOA: Multiobjective selection based on dominated hypervolume , 2007, Eur. J. Oper. Res..

[4]  Julian F. Miller,et al.  Genetic and Evolutionary Computation — GECCO 2003 , 2003, Lecture Notes in Computer Science.

[5]  Mark Fleischer,et al.  The measure of pareto optima: Applications to multi-objective metaheuristics , 2003 .

[6]  Lothar Thiele,et al.  Multiobjective evolutionary algorithms: a comparative case study and the strength Pareto approach , 1999, IEEE Trans. Evol. Comput..

[7]  R. Lyndon While,et al.  A review of multiobjective test problems and a scalable test problem toolkit , 2006, IEEE Transactions on Evolutionary Computation.

[8]  W. Rudin Principles of mathematical analysis , 1964 .

[9]  Johannes Bader,et al.  Hypervolume-based search for multiobjective optimization: Theory and methods , 2010 .

[10]  Christian Blum,et al.  Hybrid Metaheuristics , 2010, Artificial Intelligence: Foundations, Theory, and Algorithms.

[11]  Stefan Roth,et al.  Covariance Matrix Adaptation for Multi-objective Optimization , 2007, Evolutionary Computation.

[12]  Tobias Friedrich,et al.  The maximum hypervolume set yields near-optimal approximation , 2010, GECCO '10.

[13]  R. Bartle The elements of integration and Lebesgue measure , 1995 .

[14]  Gary B. Lamont,et al.  Multiobjective evolutionary algorithm test suites , 1999, SAC '99.

[15]  Frank Neumann,et al.  Multiplicative approximations and the hypervolume indicator , 2009, GECCO.

[16]  Nicola Beume,et al.  Design and Management of Complex Technical Processes and Systems by means of Computational Intelligence Methods Gradient-based / Evolutionary Relay Hybrid for Computing Pareto Front Approximations Maximizing the S-Metric , 2007 .

[17]  Lothar Thiele,et al.  The Hypervolume Indicator Revisited: On the Design of Pareto-compliant Indicators Via Weighted Integration , 2007, EMO.

[18]  Robert G. Bartle,et al.  The Elements of Integration and Lebesgue Measure: Bartle/The Elements , 1995 .

[19]  Lothar Thiele,et al.  On Set-Based Multiobjective Optimization , 2010, IEEE Transactions on Evolutionary Computation.

[20]  M. Fleischer,et al.  The Measure of Pareto Optima , 2003, EMO.

[21]  Anne Auger,et al.  Investigating and exploiting the bias of the weighted hypervolume to articulate user preferences , 2009, GECCO.

[22]  Marco Laumanns,et al.  Scalable Test Problems for Evolutionary Multiobjective Optimization , 2005, Evolutionary Multiobjective Optimization.

[23]  Eckart Zitzler,et al.  Indicator-Based Selection in Multiobjective Search , 2004, PPSN.