求解多目标优化问题的改进遗传机器学习方法  

Improved Genetic Algorithm Based on Machine Learning for Solving Multi-objective Optimization Problems

作　　者：刘明姬[1] 刘淑媛 吕显瑞[1] 

机构地区：[1]吉林大学数学学院,长春130012

出　　处：《吉林大学学报（理学版）》2006年第3期338-342,共5页Journal of Jilin University:Science Edition

摘　　要：将遗传算法与机器学习相结合,在分类器系统的基础上,引入增强因子、排挤因子、合并因子等改进因子,完善信度分配机制,提出了改进的遗传机器学习方法.并将算法应用于投资的收益与风险双目标优化模型,数值结果表明,改进算法能够寻求到数量更多、分布更广的Pareto最优解,并且具有较好的稳定性,避免了非成熟收敛.An improved genetic algorithm based on machine learning is presented. In the improved algorithm we make genetic algorithm and machine learning proceed alternatively with the assistance of improved factors. For solving multi-objective optimization problems, the improved algorithm can find more and wider Paretooptimal solutions, with more stabilitv , convergence, global search and without premature convergence.

关 键 词：遗传算法 机器学习 多目标优化 PARETO最优解 

分 类 号：O153.3[理学—数学]