一种新型的广义规则推理算法

数据挖掘是指从大型数据库的海量信息中有效进行知识发现的过程,而其效能的高低主要取决于搜索机制所依据的算法.有鉴于此,提出了一种基于个体免疫与群体进化机制于一体的一种高效的全局优化搜索算法,即基于免疫规划的广义规则推理算法.与已有算法所不同的是,广义规则推理算法不仅仅着眼于发现一些有关分类方面的信息,而是利用背景理论和先验知识在知识表示与运行效率之间相均衡的基础上,着重新知识的发现和对高级规则的预测.理论分析和仿真实验表明,广义规则推理算法有利于进化群体的相对稳定和整体性能的提高,并可以在规则提取过程中保持较高的精确度.

A Novel Algorithm for Generalized Rule Induction

Data mining usually means a process of knowledge discovery from a sea of information in a largescale database,and whose efficiency mainly depends on an algorithm used in the searching mechanism.Based on this consideration,a novel globally optimal algorithm is proposed which integrates a power of individual immunity and an evolutionary mechanism of population.Different from existing algorithms,this novel algorithm does not take great care of discovering some classifying information,but unknown knowledge or a predication on higher level rules.Theoretical analysis and simulations both show that this novel algorithm is prone to the stabilization of a population and the improvement of whole capability,and also keeping a high degree of preciseness during the rule induction.