基于粒计算的离散化算法及其应用  

Discretization Algorithm Based on Granular Computing and its Application

作　　者：史志才[1] 夏永祥[1] 周金祖[1] 

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620

出　　处：《计算机科学》2013年第06A期133-135,共3页Computer Science

基　　金：国家自然科学基金项目(61272097);上海工程技术大学学科专业建设项目(XKCZ1212);科技发展基金项目(2011XY16)资助

摘　　要：连续数值属性的离散化是粒计算理论应用的重要步骤。首先对目前的离散化算法进行了分类讨论,提出了区间粒的概念,融合熵理论定义了区间粒的粒度,进而提出了基于粒计算的连续数值属性的离散化算法,并将该算法应用于入侵检测过程;实验结果表明该算法简洁高效,能够确保入侵检测系统的检测效果。The discretization of continuous numerical attributes is an important step for the application of granular com- puting. Some current discretization algorithms are classified and discussed. The concept of section granular is proposed. Entropy theory is introduced to define the granularity of section granular. The discretization algorithm based on granular computing is proposed. This algorithm is applied to intrusion detection. The experimental results show that this algorithm is simple and effective, It can assure the accuracy of intrusion detection.

关 键 词：粒度计算 区间粒 离散化 熵 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]