The goal of linkage learning, or building block identification, is the creation of a more effective Genetic Algorithm （GA）. This paper proposes a new Linkage Learning Genetic Algorithms, named m-LLGA. With the linka...The goal of linkage learning, or building block identification, is the creation of a more effective Genetic Algorithm （GA）. This paper proposes a new Linkage Learning Genetic Algorithms, named m-LLGA. With the linkage learning module and the linkage-based genetic operation, m-LLGA is not only able to learn and record the linkage information among genes without any prior knowledge of the function being optimized. It also can use the linkage information stored in the linkage matrix to guide the selection of crossover point. The preliminary experiments on two kinds of bounded difficulty problems and a TSP problem validated the performance of m-LLGA. The m-LLGA learns the linkage of different building blocks parallel and therefore solves these problems effectively; it can also reasonably reduce the probability of building blocks being disrupted by crossover at the same time give attention to getting away from local minimum.展开更多
Evolutionary computation （EC）, a collective name rithms, is one of the fastest-growing areas in computer science. for a range of metaheuristic black-box optimization algo- Many manuals and ＂how-to＇s on the use of ...Evolutionary computation （EC）, a collective name rithms, is one of the fastest-growing areas in computer science. for a range of metaheuristic black-box optimization algo- Many manuals and ＂how-to＇s on the use of different EC methods as well as a variety of free or commercial software libraries are widely available nowadays. However, when one of these methods is applied to a real-world task, there can be many pitfalls and booby traps lurking certain aspects of the optimization problem that may lead to unsatisfactory results even if the algorithm appears to be correctly implemented and executed, These include the convergence issues, ruggedness, deceptiveness, and neutrality in the fitness landscape, epistasis, non-separability, noise leading to the need for robustness, as well as dimensionality and scalability issues, among others. In this article, we systematically discuss these related hindrances and present some possible remedies. The goal is to equip practitioners and researchers alike with a clear picture and understanding of what kind of problems can render EC applications unsuccessful and how to avoid them from the start.展开更多
基金Supported by the National Natural Science Foundation of China (No. 60234020)
文摘The goal of linkage learning, or building block identification, is the creation of a more effective Genetic Algorithm （GA）. This paper proposes a new Linkage Learning Genetic Algorithms, named m-LLGA. With the linkage learning module and the linkage-based genetic operation, m-LLGA is not only able to learn and record the linkage information among genes without any prior knowledge of the function being optimized. It also can use the linkage information stored in the linkage matrix to guide the selection of crossover point. The preliminary experiments on two kinds of bounded difficulty problems and a TSP problem validated the performance of m-LLGA. The m-LLGA learns the linkage of different building blocks parallel and therefore solves these problems effectively; it can also reasonably reduce the probability of building blocks being disrupted by crossover at the same time give attention to getting away from local minimum.
基金the National Natural Science Foundation of China under Grant Nos. U0835002, 61175065,and 61150110488the Natural Science Foundation of Anhui Province of China under Grant No. 1108085J16+2 种基金the European Union 7th Framework Program under Grant No. 247619the Chinese Academy of Sciences Fellowship for Young International Scientists underGrant No. CX05040000001Special Financial Grant from the China Postdoctoral Science Foundation under Grant No. 201104329
文摘Evolutionary computation （EC）, a collective name rithms, is one of the fastest-growing areas in computer science. for a range of metaheuristic black-box optimization algo- Many manuals and ＂how-to＇s on the use of different EC methods as well as a variety of free or commercial software libraries are widely available nowadays. However, when one of these methods is applied to a real-world task, there can be many pitfalls and booby traps lurking certain aspects of the optimization problem that may lead to unsatisfactory results even if the algorithm appears to be correctly implemented and executed, These include the convergence issues, ruggedness, deceptiveness, and neutrality in the fitness landscape, epistasis, non-separability, noise leading to the need for robustness, as well as dimensionality and scalability issues, among others. In this article, we systematically discuss these related hindrances and present some possible remedies. The goal is to equip practitioners and researchers alike with a clear picture and understanding of what kind of problems can render EC applications unsuccessful and how to avoid them from the start.
