上海交通大学学报(自然版) ›› 2012, Vol. 46 ›› Issue (02): 239-244.

• 自动化技术、计算机技术 • 上一篇    下一篇

重用抗体优良片断的免疫进化算法  

杨观赐1,2,马鑫1, 李少波1,2,钟勇2,于丽娅1   

  1. (1贵州大学 教育部现代制造技术重点实验室, 贵阳 550003;2中国科学院 成都计算机应用研究所, 成都 610041)
  • 收稿日期:2010-12-25 出版日期:2012-02-28 发布日期:2012-02-28
  • 基金资助:

    教育部新世纪优秀人才支持计划资助项目(NCET090094), 国家高技术研究发展计划(863)项目 (2009AA043203), 贵州省科学技术基金资助项目(黔科合J字[2010]2095号)

Immune Evolutionary Algorithm Reusing Excellent Genes of Antibody

 YANG  Guan-Ci-1, 2 , MA  Xin-1, LI  Shao-Bo-1, 2 , ZHONG  Yong-2, YU  Li-Ya-1   

  1. (1 Key Laboratory of Advanced Manufacturing Technology of Ministry of Education,Guizhou University, Guiyang 550003, China; 2Chengdu Institute of Computer Applications,Chinese Academy of Sciences, Chengdu 610041, China)
  • Received:2010-12-25 Online:2012-02-28 Published:2012-02-28

摘要: 基于克隆选择原理与算法,通过分析具体现象阐述了改进克隆选择算法的思想来源,设计了挖掘抗体中优秀决定基因并生成记忆集、封装优秀决定基片段、用变异抗体群中亲和度高的抗体按概率替换记忆抗体群中低亲和度抗体的方法,获得了重用抗体优良片断的克隆选择算法.借鉴强度Pareto进化算法的进化框架,提出了重用抗体优良片断的免疫进化算法.该算法通过克隆选择替代选择、交叉、重组等遗传操作.在一组0/1背包问题上的测试结果表明,所提出的算法可以有效保持种群多样性,获得较高质量的Pareto非劣解集.
 

关键词: 克隆选择, 强度帕雷托进化算法, 基因挖掘

Abstract: By expounding the ideological origin of improving the clonal selection algorithm through the analysis of the specific phenomenon, the method to extract excellent gene schema to fill a memory pool from antibody set, to package excellent gene segment, and to replace low affinity antibody with high affinity antibody with probability from mutation antibody population during updating memory antibody population was designed based on clonal selection principle and algorithm, and then an improved clonal selection algorithm reusing excellent gene segment was put forward. Refering to the framework of strength Pareto evolutionary algorithm, the immune evolutionary algorithm reusing excellent genes of antibody (RGIEM) was proposed, which implements the genetic operation such as selection, crossover and recombinant by applying the improved clonal selection algorithm. Taking a series of multiobjective 0/1 knapsack problems to check RGIEA’s performance, the results show that RGIEA is capable of maintaining the diversity of population and obtaining solutions approximating to Pareto front.

Key words: clonal selection, strength Pareto evolutionary algorithm (SPEA), gene mining

中图分类号: