ABSTRACT
There is empirical evidence that memetic algorithms (MAs) can outperform plain evolutionary algorithms (EAs). Recently the first runtime analyses have been presented proving the aforementioned conjecture rigorously by investigating Variable-Depth Search, VDS for short (Sudholt, 2008). Sudholt raised the question if there are problems where VDS performs badly. We answer this question in the affirmative in the following way. We analyze MAs with VDS, which is also known as Kernighan-Lin for the TSP, on an artificial problem and show that MAs with a simple first-improvement local search outperform VDS. Moreover, we show that the performance gap is exponential. We analyze the features leading to a failure of VDS and derive a new local search operator, coined Opportunistic Local Search, that can easily overcome regions of the search space where local optima are clustered. The power of this new operator is demonstrated on the Rastrigin function encoded for binary hypercubes. Our results provide further insight into the problem of how to prevent local search algorithms to get stuck in local optima from a theoretical perspective. The methods stem from discrete probability theory and combinatorics.
- F. Daolio, M. Tomassini, S. Vérel, and G. Ochoa. Communities of minima in local optima networks of combinatorial spaces. Physica A: Statistical Mechanics and its Applications, 390:1684--1694, 2011.Google ScholarCross Ref
- S. Droste, T. Jansen, and I. Wegener. On the analysis of the (1 + 1) evolutionary algorithm, 2002.Google Scholar
- T. Friedrich, P. S. Oliveto, D. Sudholt, and C. Witt. Theoretical analysis of diversity mechanisms for global exploration. Proceedings of the 10th annual conference on Genetic and Evolutionary Computation, 2008. Google ScholarDigital Library
- B. W. Kernighan and S. Lin. An Efficient Heuristic Procedure for Partitioning Graphs. The Bell system technical journal, 49(1):291--307, 1970.Google Scholar
- J. Kim, I. Hwang, Y.-H. Kim, and B.-R. Moon. Genetic approaches for graph partitioning: a survey. In Proceedings of the 13th annual conference on Genetic and evolutionary computation, GECCO '11, pages 473--480, New York, NY, USA, 2011. ACM. Google ScholarDigital Library
- S. Lin and B. W. Kernighan. An effective heuristic algorithm for the traveling-salesman problem. Operations Research, 21(2):498--516, 1973.Google ScholarDigital Library
- H. R. Lourenço, O. C. Martin, and T. Stützle. Iterated local search. In F. Glover and G. Kochenberger, editors, Handbook of Metaheuristics, volume 57 of International Series in Operations Research & Management Science, pages 320--353. Springer US, 2003.Google Scholar
- F. Neri, C. Cotta, and P. Moscato. Handbook of Memetic Algorithms. Studies in Computational Intelligence. Springer, 2012. Google ScholarDigital Library
- D. Sudholt. Local search in evolutionary algorithms: The impact of the local search frequency. In T. Asano, editor, ISAAC, volume 4288 of Lecture Notes in Computer Science, pages 359--368. Springer, 2006. Google ScholarDigital Library
- D. Sudholt. On the analysis of the (1 + 1) memetic algorithm. Proceedings of the 8th annual conference on Genetic and evolutionary computation - GECCO '06, page 493, 2006 Google ScholarDigital Library
- D. Sudholt. Memetic algorithms with Variable-Depth Search to overcome local optima. In Proceedings of the 10th annual conference on Genetic and evolutionary computation, pages 787--794. ACM, 2008. Google ScholarDigital Library
- D. Sudholt. Parametrization and balancing global and local search. In Handbook of Memetic Algorithms. Springer, 2012.Google Scholar
- A. Törn and A. Zhilinskas. Global optimization. Lecture notes in computer science. Springer-Verlag, 1989.Google Scholar
Index Terms
- Hybridizing evolutionary algorithms with opportunistic local search
Recommendations
Memetic algorithms with variable-depth search to overcome local optima
GECCO '08: Proceedings of the 10th annual conference on Genetic and evolutionary computationVariable-depth search (shortly VDS) is well-known as Lin-Kernighan strategy for the TSP and Kernighan-Lin for graph partitioning. The basic idea is to make a sequence of local moves and to freeze all moved combinatorial objects to prevent the search ...
Hybridizing Evolutionary Algorithms with Variable-Depth Search to Overcome Local Optima
Hybridizing evolutionary algorithms with local search has become a popular trend in recent years. There is empirical evidence for various combinatorial problems where hybrid evolutionary algorithms perform better than plain evolutionary algorithms. Due ...
Hybridizing Evolutionary Algorithms with Variable-Depth Search to Overcome Local Optima
Special Issue: Theory of Evolutionary ComputationHybridizing evolutionary algorithms with local search has become a popular trend in recent years. There is empirical evidence for various combinatorial problems where hybrid evolutionary algorithms perform better than plain evolutionary algorithms. Due ...
Comments