J4

• 数学 • 上一篇    下一篇

求解多目标优化问题的改进遗传机器学习方法

刘明姬1, 刘淑媛1,2, 吕显瑞1   

  1. 1. 吉林大学 数学学院, 长春 130012; 2. 吉林商业高等专科学校, 长春 130015
  • 收稿日期:2005-05-25 修回日期:1900-01-01 出版日期:2006-05-26 发布日期:2006-05-26
  • 通讯作者: 吕显瑞

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

LIU Ming-ji1, LIU Shu-yuan1,2, LU Xian-rui1   

  1. 1. College of Mathematics, Jilin University, Changchun 130012, China; 2. Jilin Commercial College, Changchun 130015, China
  • Received:2005-05-25 Revised:1900-01-01 Online:2006-05-26 Published:2006-05-26
  • Contact: LU Xian-rui

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.

关键词: 遗传算法, 机器学习, 多目标优化, Pareto最优解

Abstract: 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 Paretooptimal solutions, with more stability, convergence, global search and without premature convergence.

Key words: genetic algorithm, machine learning, multi-objective optimization, Pareto-optimal solution

中图分类号: 

  • O153.3