Mimetic Evolution

Abst rac t . Biological evolution is good at dealing with environmental changes: Nature ceaselessly repeats its experiments and is not misled by any explicit memory of the past. This contrasts with artificial evolution most often considering a fixed milieu, where re-generating an individual does not bring any further information. This paper aims at avoiding such uninformative operations, via some explicit memory of the past evolution: the best and the worst individuals previously met by evolution are respectively memorized within two virtual individuals. Evolution may then use these virtual individuals as social models, to be imitated or rejected. In mimetic evolution, standard crossover and mutation are replaced by a single operator, social mutation, which moves individuals farther away or closer toward the models. This new scheme involves two main parameters: the social strategy (how to move individuals with respect to the models) and the social pressure (how far the offspring go toward or away from the models). Experiments on large-sized binary problems are detailed and discussed.

[1]  Ingo Rechenberg,et al.  Evolutionsstrategie : Optimierung technischer Systeme nach Prinzipien der biologischen Evolution , 1973 .

[2]  John H. Holland,et al.  Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence , 1992 .

[3]  Hans-Paul Schwefel,et al.  Numerical optimization of computer models , 1981 .

[4]  David E. Goldberg,et al.  Genetic Algorithms in Search Optimization and Machine Learning , 1988 .

[5]  Lawrence Davis,et al.  Adapting Operator Probabilities in Genetic Algorithms , 1989, ICGA.

[6]  Gilbert Syswerda,et al.  A Study of Reproduction in Generational and Steady State Genetic Algorithms , 1990, FOGA.

[7]  John R. Koza,et al.  Genetic programming - on the programming of computers by means of natural selection , 1993, Complex adaptive systems.

[8]  David E. Goldberg,et al.  Genetic Algorithm Difficulty and the Modality of Fitness Landscapes , 1994, FOGA.

[9]  Philippe Collard,et al.  DGA: An Efficient Genetic Algorithm , 1994, ECAI.

[10]  L. C. Stayton,et al.  On the effectiveness of crossover in simulated evolutionary optimization. , 1994, Bio Systems.

[11]  John J. Grefenstette,et al.  Virtual Genetic Algorithms: First Results , 1995 .

[12]  Terry Jones,et al.  Fitness Distance Correlation as a Measure of Problem Difficulty for Genetic Algorithms , 1995, ICGA.

[13]  Terry Jones,et al.  Crossover, Macromutationand, and Population-Based Search , 1995, ICGA.

[14]  S. Baluja An Empirical Comparison of Seven Iterative and Evolutionary Function Optimization Heuristics , 1995 .

[15]  Rich Caruana,et al.  Removing the Genetics from the Standard Genetic Algorithm , 1995, ICML.

[16]  Nikolaus Hansen,et al.  On the Adaptation of Arbitrary Normal Mutation Distributions in Evolution Strategies: The Generating Set Adaptation , 1995, ICGA.

[17]  A. E. Eiben,et al.  Self-adaptivity for constraint satisfaction: learning penalty functions , 1996, Proceedings of IEEE International Conference on Evolutionary Computation.

[18]  Michèle Sebag,et al.  An Advanced Evolution Should Not Repeat its Past Errors , 1996, ICML.

[19]  Thomas Bäck,et al.  Evolutionary Algorithms in Theory and Practice , 1996 .

[20]  Thomas Bäck,et al.  Intelligent Mutation Rate Control in Canonical Genetic Algorithms , 1996, ISMIS.

[21]  Michèle Sebag,et al.  Mutation by Imitation in Boolean Evolution Strategies , 1996, PPSN.

[22]  Michèle Sebag,et al.  Toward Civilized Evolution: Developing Inhibitions , 1997, ICGA.

[23]  Isaac K. Evans,et al.  Enhancing recombination with the Complementary Surrogate Genetic Algorithm , 1997, Proceedings of 1997 IEEE International Conference on Evolutionary Computation (ICEC '97).

[24]  Marc Schoenauer,et al.  Alternative Random Initialization in Genetic Algorithms , 1997, ICGA.

[25]  D. Fogel Evolutionary algorithms in theory and practice , 1997, Complex..