项目名称: 面向进化多目标优化的局部自适应学习模型与算法研究
项目编号: No.61273317
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 公茂果
作者单位: 西安电子科技大学
项目金额: 80万元
中文摘要: 本课题针对进化多目标优化中全局搜索与局部搜索的平衡、算法参数的自适应调整、先验知识与过程知识的有效利用等关键问题展开研究。建立适用于多目标优化的拉马克学习和班德文学习两类局部学习模型,并设计两种局部学习模型的自适应协同策略。有效利用待求解问题的先验知识与搜索过程中获得的知识,建立适用于多目标优化的参数自适应学习模型,实现对高阶知识的学习。基于局部学习和自适应学习模型,提出局部搜索能力强、具备参数自适应学习能力的多目标Memetic算法,分析各学习模型对算法性能的影响,验证模型和算法的有效性与先进性。在理论研究的基础上,开展进化多目标优化在复杂网络结构分析等问题上的应用研究。研究成果在本领域重要期刊和会议发表论文15~20篇,申报国家发明专利5~6项,培养博士、硕士8~10名。
中文关键词: 进化算法;多目标优化;Memetic算法;局部搜索;网络结构分析
英文摘要: This proposal focuses on the key techniques of evolutionary multi-objective optimization in balancing the global and local search, adjusting parameters adaptively, and making full use of the priori and process knowledge. Lamarckian learning and Baldwinian learning models for multi-objective optimization will be designed. Adaptive cooperative strategy of these two local learning models will also be studied. The adaptive parameter learning model will be proposed based on problem-specific priori knowledge and process knowledge of population evolution. Based on these local learning and adaptive learning models, multi-objective memetic algorithms with enhenced local search and parameter-adptive learning will be proposed. Their effectiveness and novelties will be validated. Their applications in complex network analysis and mining will be studied. We will publish 15-20 papers in related leading journals and conferences, apply 5-6 patents, and bring up 8-10 graduate students.
英文关键词: Evolutionary Algorithm;Multi-objective Optimization;Memetic Algorithm;Local Search;Network Structural Analytics