ABSTRACT
Currently, researchers in the field of Evolutionary Algorithms (EAs) are very interested in competitions where new algorithm implementations are evaluated and compared. Usually, EA users perform their algorithm selection by following the results published in these competitions, which are typically focused on average performance measures over benchmark sets. These sets are very complete but the functions within them are usually classified into binary classes according to their separability and modality. Here we show that this binary classification could produce misleading conclusions about the performance of the EAs and, consequently, it is necessary to consider finer grained features so that better conclusions can be obtained about what scenarios are adequate or inappropriate for an EA. In particular, new elements are presented to study separability and modality in more detail than is usually done in the literature. The need for such detail in order to understand why things happen the way they do is made evident over three different EAs.
- D. H. Ackley. A connectionist machine for genetic hillclimbing. Kluwer, Boston, 1987. Google ScholarDigital Library
- E. Alba, B. Dorronsoro, M. Giacobini, and M. Tomassini. Decentralized cellular evolutionary algorithms. Handbook of Bioinspired Algorithms and Applications, 7:103--120, 2005.Google Scholar
- A. Auger and N. Hansen. A Restart CMA Evolution Strategy with Increasing Population Size. In Proceedings of the 2005 IEEE Congress on Evolutionary Computation, 2005., volume 2, pages 1769--1776, Apr 2005.Google ScholarCross Ref
- P. Caamaño, A. Prieto, J. A. Becerra, F. Bellas, and R. J. Duro. Real-valued multimodal fitness landscape characterization for evolution. In Proceedings of the 17th international conference on Neural information processing: theory and algorithms - Volume Part I, ICONIP'10, pages 567--574, Berlin, Heidelberg, 2010. Springer-Verlag. Google ScholarDigital Library
- P. Caamano, F. Bellas, J. A. Becerra, and R. J. Duro. Application domain study of evolutionary algorithms in optimization problems. In Proceedings of the 10th annual conference on Genetic and evolutionary computation, GECCO '08, pages 495--502, New York, NY, USA, 2008. ACM. Google ScholarDigital Library
- J. Garnier and L. Kallel. How to detect all maxima of a function? In Proceedings of the Second EVONET Summer School on Theoretical Aspects of Evolutionary Computing, pages 343--370, 1999. Google ScholarDigital Library
- D. E. Goldberg. Genetic Algorithms in Search, Optimization and Machine Learning. Addison-Wesley Longman Publishing Co., Inc., Boston, MA, USA, 1989. Google ScholarDigital Library
- F. Herrera, M. Lozano, and J. L. Verdegay. Tackling Real-Coded Genetic Algorithms: Operators and Tools for Behavioural Analysis. Artificial Intelligence Review, 12:265--319, 1998. Google ScholarDigital Library
- J. J. Liang, T. P. Runarsson, E. Mezura-Montes, M. Clerc, P. N. Suganthan, C. A. C. Coello, and K. Deb. Problem Definitions and Evaluation Criteria for the CEC 2006 Special Session on Constrained Real-Parameter Optimization. Technical report, Nanyang Technological University, 2006.Google Scholar
- M. Locatelli. A note on the Griewank test function. Journal of Global Optimization, 25(2):169--174, 2003. Google ScholarDigital Library
- D. Mosk-Aoyama and D. Shah. Computing separable functions via gossip. In PODC '06: Proceedings of the twenty-fifth annual ACM symposium on Principles of distributed computing, pages 113--122, New York, NY, USA, 2006. ACM. Google ScholarDigital Library
- A. Nimtawat and P. Nanakorn. Automated layout design of beam-slab floors using a genetic algorithm. Computers and Structures, 87(21-22):1308 -- 1330, 2009. Google ScholarDigital Library
- G. Rawlins. Foundations of Genetic Algorithms. Morgan Kaufmann, 1991. Google ScholarDigital Library
- J. Rönkkönen, S. Kukkonen, and K. V. Price. Real-Parameter Optimization with Differential Evolution. In The 2005 IEEE International Congress on Evolutionary Computation (CEC'05), page 12, Apr 2005.Google Scholar
- M. M. Stickberger. Genetics. Collier-McMillan Ltd, 1968.Google Scholar
- R. Storn and K. V. Price. Differential Evolution -- A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces. Journal of Global Optimization, 11:341--359, Nov 1997. Google ScholarDigital Library
- P. N. Suganthan, N. Hansen, J. J. Liang, K. Deb, Y. P. Chen, A. Auger, and S. Tiwari. Problem definitions and evaluation criteria for the CEC 2005 Special Session on Real Parameter Optimization. Technical report, Nanyang Technological University, 2005.Google Scholar
- H. Tang, S. Xue, and C. Fan. Differential evolution strategy for structural system identification. Computers and Structures, 86(21-22):2004 -- 2012, 2008. Google ScholarDigital Library
- K. Tang, X. Li, P. N. Suganthan, Z. Yang, and T. Weise. Benchmark Functions for the CEC 2010 Special Session and Competition on Large-Scale Global Optimization. Technical report, University of Science and Technology of China (USTC), School of Computer Science and Technology, Nature Inspired Computation and Applications Laboratory (NICAL): Héféi, \=Anhu\=í, China, Jan. 8, 2010.Google Scholar
- M.-H. Tseng and H.-C. Liao. The genetic algorithm for breast tumor diagnosis--The case of DNA viruses. Applied Soft Computing, 9(2):703--710, 2009. Google ScholarDigital Library
- L. D. Whitley, K. E. Mathias, S. B. Rana, and J. Dzubera. Building Better Test Functions. In Proceedings of the 6th International Conference on Genetic Algorithms, pages 239--247, San Francisco, CA, USA, 1995. Morgan Kaufmann Publishers Inc. Google ScholarDigital Library
Index Terms
- Are evolutionary algorithm competitions characterizing landscapes appropriately
Recommendations
Fitness landscapes and graphs: multimodularity, ruggedness and neutrality
GECCO '12: Proceedings of the 14th annual conference companion on Genetic and evolutionary computationOne of the most commonly-used metaphors to describe the process of heuristic search methods in solving combinatorial optimization problems is that of a fitness landscape. The landscape metaphor appears most commonly in work related to evolutionary ...
Accelerating Evolutionary Computation with Elite Obtained in Projected One-Dimensional Spaces
ICGEC '11: Proceedings of the 2011 Fifth International Conference on Genetic and Evolutionary ComputingWe propose a method for accelerating evolutionary computation (EC) searches using an elite obtained in one-dimensional space and use benchmark functions to evaluate the proposed method. The method projects individuals onto n one-dimensional spaces ...
Fitness landscapes and graphs: multimodularity, ruggedness and neutrality
GECCO '13 Companion: Proceedings of the 15th annual conference companion on Genetic and evolutionary computationOne of the most commonly-used metaphors to describe the process of heuristic search methods in solving combinatorial optimization problems is that of a fitness landscape. The landscape metaphor appears most commonly in work related to evolutionary ...
Comments