A general technique for detecting equilibria in finite non cooperative games is proposed. Fundamental idea is that every equilibrium is characterized by a binary relation on the game strategies. This relation - called generative relation -- induces an appropriate domination concept. Game equilibrium is described as the set of non dominated strategies with respect to the generative relation. Slight generalizations of some well known equilibrium concepts are proposed. A population of strategies is evolved according to a domination-based ranking in oder to produce better and better equilibrium approximations. Eventually the process converges towards the game equilibrium. The proposed technique opens an way for qualitative approach of game equilibria. In order to illustrate the proposed evolutionary technique different equilibria for different continuous games are studied. Numerical experiments indicate the potential of the proposed concepts and technique.
{"title":"Generative relations for evolutionary equilibria detection","authors":"D. Dumitrescu, R. Lung, T. Mihoc","doi":"10.1145/1569901.1570103","DOIUrl":"https://doi.org/10.1145/1569901.1570103","url":null,"abstract":"A general technique for detecting equilibria in finite non cooperative games is proposed. Fundamental idea is that every equilibrium is characterized by a binary relation on the game strategies. This relation - called generative relation -- induces an appropriate domination concept. Game equilibrium is described as the set of non dominated strategies with respect to the generative relation. Slight generalizations of some well known equilibrium concepts are proposed. A population of strategies is evolved according to a domination-based ranking in oder to produce better and better equilibrium approximations. Eventually the process converges towards the game equilibrium. The proposed technique opens an way for qualitative approach of game equilibria. In order to illustrate the proposed evolutionary technique different equilibria for different continuous games are studied. Numerical experiments indicate the potential of the proposed concepts and technique.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115607070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
C. Müller, B. Baumgartner, Georg Ofenbeck, B. Schrader, I. Sbalzarini
We present pCMALib, a parallel software library that implements the Evolution Strategy with Covariance Matrix Adaptation (CMA-ES). The library is written in Fortran 90/95 and uses the Message Passing Interface (MPI) for efficient parallelization on shared and distributed memory machines. It allows single CMA-ES optimization runs, embarrassingly parallel CMA-ES runs, and coupled parallel CMA-ES runs using a cooperative island model. As one instance of an island model CMA-ES, the recently presented Particle Swarm CMA-ES (PS-CMA-ES) is included using collaborative concepts from Swarm Intelligence for the migration model. Special attention has been given to an efficient design of the MPI communication protocol, a modular software architecture, and a user-friendly programming interface. The library includes a Matlab interface and is supplemented with an efficient Fortran implementation of the official CEC 2005 set of 25 real-valued benchmark functions. This is the first freely available Fortran implementation of this standard benchmark test suite. We present test runs and parallel scaling benchmarks on Linux clusters and multi-core desktop computers, showing good parallel efficiencies and superior computational performance compared to the reference implementation.
{"title":"pCMALib: a parallel fortran 90 library for the evolution strategy with covariance matrix adaptation","authors":"C. Müller, B. Baumgartner, Georg Ofenbeck, B. Schrader, I. Sbalzarini","doi":"10.1145/1569901.1570090","DOIUrl":"https://doi.org/10.1145/1569901.1570090","url":null,"abstract":"We present pCMALib, a parallel software library that implements the Evolution Strategy with Covariance Matrix Adaptation (CMA-ES). The library is written in Fortran 90/95 and uses the Message Passing Interface (MPI) for efficient parallelization on shared and distributed memory machines. It allows single CMA-ES optimization runs, embarrassingly parallel CMA-ES runs, and coupled parallel CMA-ES runs using a cooperative island model. As one instance of an island model CMA-ES, the recently presented Particle Swarm CMA-ES (PS-CMA-ES) is included using collaborative concepts from Swarm Intelligence for the migration model. Special attention has been given to an efficient design of the MPI communication protocol, a modular software architecture, and a user-friendly programming interface. The library includes a Matlab interface and is supplemented with an efficient Fortran implementation of the official CEC 2005 set of 25 real-valued benchmark functions. This is the first freely available Fortran implementation of this standard benchmark test suite. We present test runs and parallel scaling benchmarks on Linux clusters and multi-core desktop computers, showing good parallel efficiencies and superior computational performance compared to the reference implementation.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115769238","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Bloat may occur when evolution allows chromosome growth. Recently it has been shown that elitism can inhibit bloat. Here we study interactions between growth, elitism, and fitness landscapes. Our results show that in some cases elitism neither constrains growth nor increases the rate of fitness accumulation, and when elitism does constrain growth it may stall the search completely.
{"title":"Elitism, fitness, and growth","authors":"Gerardo Gonzalez, Dean Frederick Hougen","doi":"10.1145/1569901.1570199","DOIUrl":"https://doi.org/10.1145/1569901.1570199","url":null,"abstract":"Bloat may occur when evolution allows chromosome growth. Recently it has been shown that elitism can inhibit bloat. Here we study interactions between growth, elitism, and fitness landscapes. Our results show that in some cases elitism neither constrains growth nor increases the rate of fitness accumulation, and when elitism does constrain growth it may stall the search completely.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"133 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127237288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The paper presents a new optimization algorithm, which adapts the behavior of honey bees during their search for nectar. In addition to the established ant algorithms, bee-inspired algorithms represent a relatively young form of solution procedures, whose applicability to the solution of complex optimization problems has already been shown. The herein presented two-stage approach belongs to the class of metaheuristics to control a construction heuristic and has been applied successfully to the NP-hard Vehicle Routing Problem with Time Windows (VRPTW). Within the paper, evaluation results are presented, which compare the developed algorithm to some of the most successful procedures for the solution of benchmark problems. The pursued approach gives the best results so far for a metaheuristic to control a construction heuristic.
{"title":"The bee colony-inspired algorithm (BCiA): a two-stage approach for solving the vehicle routing problem with time windows","authors":"S. Häckel, P. Dippold","doi":"10.1145/1569901.1569906","DOIUrl":"https://doi.org/10.1145/1569901.1569906","url":null,"abstract":"The paper presents a new optimization algorithm, which adapts the behavior of honey bees during their search for nectar. In addition to the established ant algorithms, bee-inspired algorithms represent a relatively young form of solution procedures, whose applicability to the solution of complex optimization problems has already been shown. The herein presented two-stage approach belongs to the class of metaheuristics to control a construction heuristic and has been applied successfully to the NP-hard Vehicle Routing Problem with Time Windows (VRPTW). Within the paper, evaluation results are presented, which compare the developed algorithm to some of the most successful procedures for the solution of benchmark problems. The pursued approach gives the best results so far for a metaheuristic to control a construction heuristic.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127326289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper describes an innovative analog circuit design optimization kernel. The new approach generates fuzzy models for qualitative reasoning based on a DOE approach. The models are then used within a standard genetic algorithm implementation enhancing the search by incorporating design knowledge represented by the fuzzy models. The achieved performance is discussed for a set of well known analog circuit structures.
{"title":"FUGA: a fuzzy-genetic analog circuit optimization kernel","authors":"Pedro Sousa, C. Duarte, N. Horta","doi":"10.1145/1569901.1570156","DOIUrl":"https://doi.org/10.1145/1569901.1570156","url":null,"abstract":"This paper describes an innovative analog circuit design optimization kernel. The new approach generates fuzzy models for qualitative reasoning based on a DOE approach. The models are then used within a standard genetic algorithm implementation enhancing the search by incorporating design knowledge represented by the fuzzy models. The achieved performance is discussed for a set of well known analog circuit structures.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"74 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124726701","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Benchmarking of a team based model of Genetic Programming demonstrates that the naturally embedded style of feature selection is usefully extended by the teaming metaphor to provide solutions in terms of exceptionally low attribute counts. To take this concept to its logical conclusion the teaming model must be able to build teams with a non-overlapping behavioral trait, from a single population. The Symbiotic Bid-Based (SBB) algorithm is demonstrated to fit this purpose under an evaluation utilizing data sets with 650 to 5,000 attributes. The resulting solutions are one to two orders simpler than solutions identified under the alternative embedded paradigms of C4.5 and MaxEnt.
{"title":"Benchmarking coevolutionary teaming under classification problems with large attribute spaces","authors":"J. Doucette, P. Lichodzijewski, M. Heywood","doi":"10.1145/1569901.1570226","DOIUrl":"https://doi.org/10.1145/1569901.1570226","url":null,"abstract":"Benchmarking of a team based model of Genetic Programming demonstrates that the naturally embedded style of feature selection is usefully extended by the teaming metaphor to provide solutions in terms of exceptionally low attribute counts. To take this concept to its logical conclusion the teaming model must be able to build teams with a non-overlapping behavioral trait, from a single population. The Symbiotic Bid-Based (SBB) algorithm is demonstrated to fit this purpose under an evaluation utilizing data sets with 650 to 5,000 attributes. The resulting solutions are one to two orders simpler than solutions identified under the alternative embedded paradigms of C4.5 and MaxEnt.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125806610","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Luis Martí, Jesús García, A. Berlanga, J. M. Molina
The multi-objective optimization neural estimation of distribution algorithm (MONEDA) was devised with the purpose of dealing with the model-building issues of MOEDAs and, therefore address their scalability. In this paper we put forward a comprehensive set of experiments that intends to compare MONEDA with similar approaches when solving complex community accepted MOPs. In particular, we deal with the Walking Fish Group scalable test problem set (WFG). These tests aim to establish the optimizing capacity of MONEDA and the consistency as an optimization method. The fundamental conclusion of these assessment is that we provide strong evidences of the viability of MONEDA for handling hard and complex high-dimensional problems and its superior performance when compared to similar approaches. In spite of the fact that obviously further studies are necessary, these extensive experiments have provided solid ground for the use of MONEDA in more ambitious real-world applications.
针对多目标优化分布神经估计算法(MONEDA)的建模问题，提出了多目标优化分布神经估计算法(MONEDA)，从而解决了其可扩展性问题。在本文中，我们提出了一套全面的实验，旨在将MONEDA与类似方法在解决复杂的社区接受的MOPs时进行比较。特别地，我们处理了Walking Fish Group可伸缩测试问题集(WFG)。这些试验旨在建立MONEDA的优化能力和一致性作为一种优化方法。这些评估的基本结论是，我们提供了强有力的证据，证明MONEDA在处理困难和复杂的高维问题方面的可行性，以及与类似方法相比其优越的性能。尽管显然需要进一步的研究，但这些广泛的实验为MONEDA在更雄心勃勃的实际应用中使用提供了坚实的基础。
{"title":"Solving complex high-dimensional problems with the multi-objective neural estimation of distribution algorithm","authors":"Luis Martí, Jesús García, A. Berlanga, J. M. Molina","doi":"10.1145/1569901.1569987","DOIUrl":"https://doi.org/10.1145/1569901.1569987","url":null,"abstract":"The multi-objective optimization neural estimation of distribution algorithm (MONEDA) was devised with the purpose of dealing with the model-building issues of MOEDAs and, therefore address their scalability. In this paper we put forward a comprehensive set of experiments that intends to compare MONEDA with similar approaches when solving complex community accepted MOPs. In particular, we deal with the Walking Fish Group scalable test problem set (WFG). These tests aim to establish the optimizing capacity of MONEDA and the consistency as an optimization method. The fundamental conclusion of these assessment is that we provide strong evidences of the viability of MONEDA for handling hard and complex high-dimensional problems and its superior performance when compared to similar approaches. In spite of the fact that obviously further studies are necessary, these extensive experiments have provided solid ground for the use of MONEDA in more ambitious real-world applications.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123742849","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, we rigorously analyse how the magnitude and frequency of change may affect the performance of the algorithm (1+1) EAdyn on a set of artificially designed pseudo-Boolean functions, given a simple but well-defined dynamic framework. We demonstrate some counter-intuitive scenarios that allow us to gain a better understanding of how the dynamics of a function may affect the runtime of an algorithm. In particular, we present the function Magnitude, where the time it takes for the (1+1) EAdyn to relocate the global optimum is less than n2log n (i.e., efficient) with overwhelming probability if the magnitude of change is large. For small changes of magnitude, on the other hand, the expected time to relocate the global optimum is eΩ(n) (i.e., highly inefficient). Similarly, the expected runtime of the (1+1) EAdyn on the function Balance is O(n2) (efficient) for a high frequencies of change and nΩ(√n) (highly inefficient) for low frequencies of change. These results contribute towards a better understanding of dynamic optimisation problems in general and show how traditional analytical methods may be applied in the dynamic case.
{"title":"Dynamic evolutionary optimisation: an analysis of frequency and magnitude of change","authors":"Philipp Rohlfshagen, P. Lehre, X. Yao","doi":"10.1145/1569901.1570131","DOIUrl":"https://doi.org/10.1145/1569901.1570131","url":null,"abstract":"In this paper, we rigorously analyse how the magnitude and frequency of change may affect the performance of the algorithm (1+1) EAdyn on a set of artificially designed pseudo-Boolean functions, given a simple but well-defined dynamic framework. We demonstrate some counter-intuitive scenarios that allow us to gain a better understanding of how the dynamics of a function may affect the runtime of an algorithm. In particular, we present the function Magnitude, where the time it takes for the (1+1) EAdyn to relocate the global optimum is less than n2log n (i.e., efficient) with overwhelming probability if the magnitude of change is large. For small changes of magnitude, on the other hand, the expected time to relocate the global optimum is eΩ(n) (i.e., highly inefficient). Similarly, the expected runtime of the (1+1) EAdyn on the function Balance is O(n2) (efficient) for a high frequencies of change and nΩ(√n) (highly inefficient) for low frequencies of change. These results contribute towards a better understanding of dynamic optimisation problems in general and show how traditional analytical methods may be applied in the dynamic case.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122813302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Patrick O. Stalph, Martin Volker Butz, D. Goldberg, Xavier Llorà
Many successful applications have proven the potential of Learning Classifier Systems and the XCS classifier system in particular in datamining, reinforcement learning, and function approximation tasks. Recent research has shown that XCS is a highly flexible system, which can be adapted to the task at hand by adjusting its condition structures, learning operators, and prediction mechanisms. However, fundamental theory concerning the scalability of XCS dependent on these enhancements and problem difficulty is still rather sparse and mainly restricted to boolean function problems. In this article we developed a learning scalability theory for XCSF---the XCS system applied to real-valued function approximation problems. We determine crucial dependencies on functional properties and on the developed solution representation and derive a theoretical scalability model out of these constraints. The theoretical model is verified with empirical evidence. That is, we show that given a particular problem difficulty and particular representational constraints XCSF scales optimally. In consequence, we discuss the importance of appropriate prediction and condition structures regarding a given problem and show that scalability properties can be improved by polynomial orders, given an appropriate, problem-suitable representation.
{"title":"On the scalability of XCS(F)","authors":"Patrick O. Stalph, Martin Volker Butz, D. Goldberg, Xavier Llorà","doi":"10.1145/1569901.1570077","DOIUrl":"https://doi.org/10.1145/1569901.1570077","url":null,"abstract":"Many successful applications have proven the potential of Learning Classifier Systems and the XCS classifier system in particular in datamining, reinforcement learning, and function approximation tasks. Recent research has shown that XCS is a highly flexible system, which can be adapted to the task at hand by adjusting its condition structures, learning operators, and prediction mechanisms. However, fundamental theory concerning the scalability of XCS dependent on these enhancements and problem difficulty is still rather sparse and mainly restricted to boolean function problems. In this article we developed a learning scalability theory for XCSF---the XCS system applied to real-valued function approximation problems. We determine crucial dependencies on functional properties and on the developed solution representation and derive a theoretical scalability model out of these constraints. The theoretical model is verified with empirical evidence. That is, we show that given a particular problem difficulty and particular representational constraints XCSF scales optimally. In consequence, we discuss the importance of appropriate prediction and condition structures regarding a given problem and show that scalability properties can be improved by polynomial orders, given an appropriate, problem-suitable representation.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122871943","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper demonstrates the use of genetic algorithms for evolving a grandmaster-level evaluation function for a chess program. This is achieved by combining supervised and unsupervised learning. In the supervised learning phase the organisms are evolved to mimic the behavior of human grandmasters, and in the unsupervised learning phase these evolved organisms are further improved upon by means of coevolution. While past attempts succeeded in creating a grandmaster-level program by mimicking the behavior of existing computer chess programs, this paper presents the first successful attempt at evolving a state-of-the-art evaluation function by learning only from databases of games played by humans. Our results demonstrate that the evolved program outperforms a two-time World Computer Chess Champion.
{"title":"Simulating human grandmasters: evolution and coevolution of evaluation functions","authors":"E. David, H. J. Herik, Moshe Koppel, N. Netanyahu","doi":"10.1145/1569901.1570100","DOIUrl":"https://doi.org/10.1145/1569901.1570100","url":null,"abstract":"This paper demonstrates the use of genetic algorithms for evolving a grandmaster-level evaluation function for a chess program. This is achieved by combining supervised and unsupervised learning. In the supervised learning phase the organisms are evolved to mimic the behavior of human grandmasters, and in the unsupervised learning phase these evolved organisms are further improved upon by means of coevolution. While past attempts succeeded in creating a grandmaster-level program by mimicking the behavior of existing computer chess programs, this paper presents the first successful attempt at evolving a state-of-the-art evaluation function by learning only from databases of games played by humans. Our results demonstrate that the evolved program outperforms a two-time World Computer Chess Champion.","PeriodicalId":193093,"journal":{"name":"Proceedings of the 11th Annual conference on Genetic and evolutionary computation","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114188505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：