ABSTRACT
Black-box optimization of a previously unknown problem can often prove to be a demanding task. In order for the optimization process to be as efficient as possible, one must first recognize the nature of the problem at hand and then proceed to choose the algorithm exhibiting the best performance for that type of problem. The problem characterization is done via underlying fitness landscape features, which allow to identify similarities and differences between various problems.
In this paper we present first steps towards an adaptive landscape analysis. Our approach is aimed at taking a closer look into how features evolve during the optimization process and whether this information can be used to discriminate between different problems. The motivation of our work is to understand if and how one could exploit the information provided by the features to improve on dynamic algorithm selection and configuration. Put differently, our goal is to leverage landscape analysis to adjust the choice of the algorithm on the fly, i.e., during the optimization process itself.
- Nacim Belkhir, Johann Dreo, Pierre Saveant, and Marc Schoenauer. 2017. Per instance algorithm configuration of CMA-ES with limited budget. In Proc. of Genetic and Evolutionary Computation Conference (GECCO'17). ACM, 681--688. Google ScholarDigital Library
- Edmund K. Burke, Michel Gendreau, Matthew R. Hyde, Graham Kendall, Gabriela Ochoa, Ender Ozcan, and Rong Qu. 2013. Hyper-heuristics: a survey of the state of the art. JORS 64, 12 (2013), 1695--1724.Google ScholarCross Ref
- Benjamin Doerr and Carola Doerr. 2018. Theory of Parameter Control for Discrete Black-Box Optimization: Provable Performance Gains Through Dynamic Parameter Choices. CoRR abs/1804.05650 (2018). arXiv:1804.05650Google Scholar
- Michael G. Epitropakis and Edmund K. Burke. 2018. Hyper-heuristics. In Handbook of Heuristics. Springer, 489--545.Google Scholar
- Nikolaus Hansen, Youhei Akimoto, and Petr Baudis. 2019. CMA-ES/pycma on Github. Zenodo, (Feb. 2019).Google Scholar
- Nikolaus Hansen, Anne Auger, Olaf Mersmann, Tea Tusar, and Dimo Brockhoff. 2016. COCO: A Platform for Comparing Continuous Optimizers in a Black-Box Setting. CoRR abs/1603.08785 (2016). arXiv:1603.08785Google Scholar
- Nikolaus Hansen and Andreas Ostermeier. 2001. Completely Derandomized Self-Adaptation in Evolution Strategies. Evolutionary Computation 9, 2 (2001), 159--195. Google ScholarDigital Library
- Giorgos Karafotias, Mark Hoogendoorn, and Ágoston E. Eiben. 2015. Parameter Control in Evolutionary Algorithms: Trends and Challenges. IEEE Trans. Evolutionary Computation 19, 2 (2015), 167--187.Google ScholarCross Ref
- Pascal Kerschke, Mike Preuss, Simon Wessing, and Heike Trautmann. 2015. Detecting Funnel Structures by Means of Exploratory Landscape Analysis. In Proc, of Genetic and Evolutionary Computation Conference (GECCO'15). ACM, New York, NY, USA, 265--272. Google ScholarDigital Library
- Pascal Kerschke and Heike Trautmann. 2016. The R-Package FLACCO for exploratory landscape analysis with applications to multi-objective optimization problems. In Proc. of IEEE Congress on Evolutionary Computation (CEC'16). IEEE, 5262--5269.Google ScholarDigital Library
- Monte Lunacek and Darrell Whitley. 2006. The Dispersion Metric and the CMA Evolution Strategy. In Proc. of Genetic and Evolutionary Computation Conference (GECCO'06). ACM, 477--484. Google ScholarDigital Library
- Olaf Mersmann, Bernd Bischl, Heike Trautmann, Mike Preuss, Claus Weihs, and Günter Rudolph. 2011. Exploratory landscape analysis. In Proc. of Genetic and Evolutionary Computation Conference (GECCO'11). ACM, 829--836. Google ScholarDigital Library
- Olaf Mersmann, Mike Preuss, and Heike Trautmann. 2010. Benchmarking Evolutionary Algorithms: Towards Exploratory Landscape Analysis. In Proc. of Parallel Problem Solving from Nature (PPSN'10). Springer, 73--82. Google ScholarDigital Library
- Mario A. Muñoz, Michael Kirley, and Saman K. Halgamuge. 2015. Exploratory Landscape Analysis of Continuous Space Optimization Problems Using Information Content. IEEE Trans. Evolutionary Computation 19, 1 (2015), 74--87.Google ScholarCross Ref
- Sander van Rijn, Carola Doerr, and Thomas Bäck. 2018. Towards an Adaptive CMA-ES Configurator. In Proc. of Parallel Problem Solving from Nature (PPSN'18). Springer, 54--65.Google ScholarCross Ref
- Sander van Rijn, Hao Wang, Matthijs van Leeuwen, and Thomas Bäck. 2016. Evolving the structure of Evolution Strategies. In Proc. of IEEE Symposium Series on Computational Intelligence (SSCI'16). IEEE, 1--8.Google ScholarCross Ref
- Diederick Vermetten, Sander van Rijn, Thomas Bäck, and Carola Doerr. 2019. Online Selection of CMA-ES Variants. In Proc. of Genetic and Evolutionary Computation Conference (GECCO'19). ACM. To appear Google ScholarDigital Library
- David H. Wolpert and William G. Macready. 1997. No free lunch theorems for optimization. IEEE Trans. Evolutionary Computation 1, 1 (1997), 67--82. Google ScholarDigital Library
Index Terms
- Adaptive landscape analysis
Recommendations
Bag of local landscape features for fitness landscape analysis
Evolutionary computation is a research field dealing with black-box and complex optimization problems whose fitness landscapes are usually unknown in advance. It is difficult to select an appropriate evolutionary algorithm and parameters for a given ...
The landscape adaptive particle swarm optimizer
Several modified particle swarm optimizers are proposed in this paper. In DVPSO, a distribution vector is used in the update of velocity. This vector is adjusted automatically according to the distribution of particles in each dimension. In COPSO, the ...
Local Landscape Patterns for Fitness Landscape Analysis
Simulated Evolution and LearningAbstractAlmost all problems targeted by evolutionary computation are black-box or heavily complex, and their fitness landscapes usually are unknown. Selection of the appropriate search algorithm and parameters is a crucial topic when the landscape of a ...
Comments