计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (20): 210-212.DOI: 10.3778/j.issn.1002-8331.2010.20.057

• 人工智能 • 上一篇    下一篇

遗传-粒子群的投影寻踪模型

万中英1,廖海波2,王明文1   

  1. 1.江西师范大学 计算机信息工程学院,南昌 330022
    2.江西师范大学 科学技术学院,南昌 330027
  • 收稿日期:2010-04-15 修回日期:2010-05-18 出版日期:2010-07-11 发布日期:2010-07-11
  • 通讯作者: 万中英

Projection pursuit model using genetic-PSO

WAN Zhong-ying1,LIAO Hai-bo2,WANG Ming-wen1   

  1. 1.School of Computer Information Engineering,Jiangxi Normal University,Nanchang 330022,China
    2.School of Science and Technology,Jiangxi Normal University,Nanchang 330027,China
  • Received:2010-04-15 Revised:2010-05-18 Online:2010-07-11 Published:2010-07-11
  • Contact: WAN Zhong-ying

摘要: 以前的投影寻踪研究都是采用遗传算法来寻找最优的投影方向,但遗传算法对初始种群的选择有一定的依赖性,收敛速度较慢,而且得到的也未必是最优解。粒子群算法是一种模拟鸟群飞行觅食的行为,通过个体之间的协作来寻找最优解的进化计算技术。根据遗传算法和粒子群算法的优缺点,将两者有效地结合在一起,提出了遗传-粒子群的投影寻踪模型。该方法能有效地解决投影寻踪模型中投影方向的寻优问题,并将该方法应用于文本分类,在Reuters-21578文档集上分别采用KNN和朴素贝叶斯方法进行实验,结果表明此方法能有效提取投影方向,取得了满意的分类效果,也提高了算法收敛到最优解的能力。

关键词: 遗传算法, 粒子群算法, 投影方向, 投影寻踪, 文本分类

Abstract: The previous studies are based on projection pursuit of genetic algorithm to find the optimal projection direction,but the genetic algorithm to the choice of the initial population has a certain dependence,the convergence rate is slow,and the solution obtained may not be optimal solution.Particle swarm algorithm which is a kind of swarm intelligent optimization algorithms,is put forward by foraging act of research and observation about groups of birds.This paper proposes the projection pursuit model using genetic-particle swarm optimization according to the advantages and disadvantages of genetic algorithms and particle swarm optimization.This method can effectively solve the optimization problem of the projection direction.The paper applies this method in the text classification,in the Reuters-21578 document sets KNN and the Naive Bayesian are used,experimental results show that this method can effectively extract the projection direction to obtain a satisfactory classification results,and improve algorithm convergence to the optimal solution capability.

Key words: genetic algorithm, Partical Swarm Optimization(PSO), projection direction, projection pursuit, text classification

中图分类号: