Hypervolume Maximization via Set Based Newton’s Method
The hypervolume indicator is one of the most widely used tool to measure the performance in evolutionary multi-objective optimization. While derivative free methods such as specialized evolutionary algorithms received considerable attention in the past, the investigation of derivative based methods is still scarce. In this work, we aim to make a contribution to fill this gap.
Based on the hypervolume gradient that has recently been proposed for general unconstrained multi-objective optimization problems, we first investigate the behavior of the related hypervolume flow. Under this flow, populations evolve toward a final state (population) whose hypervolume indicator is locally maximal. Some insights obtained on selected test functions explain to a certain extend observations made in previous studies and give some possible insights into the application of mathematical programming techniques to this problem. Further, we apply a population-based version of the Newton Raphson method for the maximization of the hypervolume. Fast set-based convergence can be observed towards optimal populations, however, the results indicate that the success depends crucially on the choice of the initial population.
Keywordsmulti-objective optimization hypervolume indicator set based optimization multi-objective gradient Newton Raphson method
Unable to display preview. Download preview PDF.
- 3.Zitzler, E.: Evolutionary Algorithms for Multiobjective Optimization: Methods and Applications. PhD thesis, ETH Zurich, Switzerland (1999)Google Scholar
- 7.Koch, P., Kramer, O., Rudolph, G., Beume, N.: On the hybridization of sms-emoa and local search for continuous multiobjective optimization. In: Proceedings of the 11th Annual Conference on Genetic and Evolutionary Computation, pp. 603–610. ACM (2009)Google Scholar
- 9.Emmerich, M., Deutz, A., Beume, N.: Gradient-based/evolutionary relay hybrid for computing pareto front approximations maximizing the s-metric. In: Bartz-Beielstein, T., Blesa Aguilera, M.J., Blum, C., Naujoks, B., Roli, A., Rudolph, G., Sampels, M. (eds.) HCI/ICCV 2007. LNCS, vol. 4771, pp. 140–156. Springer, Heidelberg (2007)CrossRefGoogle Scholar
- 10.Emmerich, M., Deutz, A.: Time complexity and zeros of the hypervolume indicator gradient field. In: Schuetze, O., Coello, C.A., Tantar, A.-A., Tantar, E., Bouvry, P., Moral, P.D., Legrand, P. (eds.) EVOLVE - A Bridge between Probability, Set Oriented Numerics, and Evolutionary Computation III. SCI, vol. 500, pp. 169–193. Springer, Heidelberg (2014)CrossRefGoogle Scholar
- 11.Hernández, V.A.S., Schütze, O., Rudolph, G., Trautmann, H.: The directed search method for pareto front approximations with maximum dominated hypervolume. In: Emmerich, M., et al. (eds.) EVOLVE - A Bridge between Probability, Set Oriented Numerics,and Evolutionary Computation IV. AISC, vol. 227, pp. 189–205. Springer, Heidelberg (2013)CrossRefGoogle Scholar
- 12.Pareto, V.: Manual of Political Economy. The MacMillan Press (1971 (original edition in French in 1927))Google Scholar
- 14.Auger, A., Bader, J., Brockhoff, D., Zitzler, E.: Theory of the hypervolume indicator: optimal mu-distributions and the choice of the reference point. In: FOGA 2009: Proceedings of the Tenth ACM SIGEVO Workshop on Foundations of Genetic Algorithms, pp. 87–102. ACM, New York (2009)CrossRefGoogle Scholar
- 15.Emmerich, M., Deutz, A. (eds.): EVOLVE 2013 - A Bridge Between Probability. Set Oriented Numerics, and Evolutionary Computation - Short Paper and Extended Abstract Proceedings, LIACS, Leiden University (July 2013)Google Scholar
- 16.Emmerich, M., Deutz, A.: A family of test problems with pareto-fronts of variable curvature based on super-spheres. In: MCDM 2006, Chania, Greece, June 19-23 (2006)Google Scholar
- 17.Schütze, O.: Set Oriented Methods for Global Optimization. PhD thesis, University of Paderborn (2004), http://ubdata.uni-paderborn.de/ediss/17/2004/schuetze/
- 19.Witting, K.: Numerical Algorithms for the Treatment of Parametric Optimization Problems and Applications. PhD thesis, University of Paderborn (2012)Google Scholar