交互式遗传算法中用户的认知规律及其应用

发布时间：2018-06-30 19:33

本文选题：遗传算法 + 交互　；参考：《中国矿业大学》2009年博士论文

【摘要】： 交互式遗传算法把人的智慧和遗传算法结合起来,主要用于解决无法建立显式函数的隐式性能指标优化问题。交互式遗传算法在发挥人类智慧的同时,也需要面对人自身的局限性。人的认知局限性和易疲劳特点,使得交互式遗传算法的种群规模较小和进化代数较少,这限制了交互式遗传算法的优化性能。许多学者研究了改进交互式遗传算法性能的方法,这些方法几乎都与用户偏好信息相关。由于用户偏好信息往往综合了多种用户认知规律,因此,为了更好地获取用户偏好信息,必须深入研究交互式遗传算法中用户的认知规律。但是,已有研究成果中对用户认知规律的研究却很少。本文通过研究交互式遗传算法中用户的认知规律,进而研究交互式遗传算法收敛理论和性能改进方法。本文内容主要从以下5个方面展开:(1)研究交互式遗传算法中用户的参照认知规律,分别考虑理论参照认知和实际参照认知的算法收敛理论,提出交互式遗传算法全局收敛的强条件和弱条件;(2)研究交互式遗传算法中用户的理性认知规律,提出用户保持理性是交互式遗传算法全局收敛的充分条件,并针对赋予适应值的不同方法给出用户保持理性的最大进化代数估计;(3)研究交互式遗传算法中用户的不确定性认知规律,给出用户偏好知识提取、表示及更新方法,并结合定向变异,提出了改进算法性能的方法;(4)研究交互式遗传算法中用户的选择性注意认知规律,提出获取用户选择性注意的种群初始化方法和跟踪用户选择性注意的个体生成方法,并给合用户选择性注意知识,提出算法性能改进的方法;(5)研究交互式遗传算法系统的实现,给出交互式遗传算法的系统实现框架、模块划分,并给出基于交互式遗传算法的三维动漫人物造型系统。本文的研究成果不仅丰富了交互式遗传算法的基础理论,而且为把交互式遗传算法应用于工程实践提供了理论指导。
[Abstract]:Interactive genetic algorithm combines human intelligence with genetic algorithm, which is mainly used to solve the problem of implicit performance index optimization which can not establish explicit function. Interactive genetic algorithm not only exerts human intelligence, but also faces human limitations. Due to the cognitive limitation and fatigue, the interactive genetic algorithm has smaller population size and less evolutionary algebra, which limits the optimization performance of the interactive genetic algorithm. Many scholars have studied methods to improve the performance of interactive genetic algorithms, which are almost related to user preference information. Because user preference information often synthesizes a variety of user cognitive laws, in order to obtain user preference information better, it is necessary to deeply study the cognitive law of users in interactive genetic algorithm (IGA). However, there are few researches on the law of user cognition in the existing research results. In this paper, the convergence theory and performance improvement method of interactive genetic algorithm are studied by studying the cognitive law of users in interactive genetic algorithm. The main contents of this paper are as follows: (1) studying the rules of user's reference cognition in interactive genetic algorithm, considering the convergence theory of theoretical reference cognition and practical reference cognition, respectively. The strong and weak conditions for the global convergence of interactive genetic algorithms are proposed. (2) the rational cognitive laws of users in interactive genetic algorithms are studied, and the sufficient conditions for global convergence of interactive genetic algorithms are proposed. The maximum evolutionary algebraic estimation of user's rationality is given according to the different methods of endowing fitness. (3) the uncertain cognition law of user in interactive genetic algorithm is studied, and the method of extracting, expressing and updating user's preference knowledge is given. Combined with directional mutation, a method to improve the performance of the algorithm is proposed. (4) the rules of selective attention cognition of users in interactive genetic algorithm are studied. This paper proposes a population initialization method for obtaining user selective attention and an individual generation method for tracking user selective attention, and proposes a method to improve the performance of the algorithm by combining the knowledge of user selective attention. (5) the realization of interactive genetic algorithm system is studied. The system implementation framework and module partition of interactive genetic algorithm are presented, and the 3D animation character modeling system based on interactive genetic algorithm is presented. The research results of this paper not only enrich the basic theory of interactive genetic algorithm, but also provide theoretical guidance for the application of interactive genetic algorithm in engineering practice.
【学位授予单位】：中国矿业大学
【学位级别】：博士
【学位授予年份】：2009
【分类号】：TP18

【引证文献】