欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
传统的模糊支持向量机隶属度函数是基于样本点到类中心点的距离进行设计的，这对非规则形状分布数据很不合理。在基于粗糙集和支持向量机建立入侵检测模型里，使用粗糙集理论挖掘出各条件属性对决策属性的影响决策程度，提出基于样本点与类中心点属性比较加权的新隶属度函数构造方法。该方法用于此模型，可以有效降低隶属度函数对样本集几何形状的依赖，能够有效地区分样本点、噪音点以及孤立点。实验表明，与支持向量机和传统基于类中心距离的模糊支持向量机相比，新的基于属性相关的隶属度函数的模糊支持向量机达到最好的分类效果，而且新隶属度方法简单易行，运行速度快。
英文摘要：
The traditional fuzzy membership function of support vector machine is based on the distance between sample and the class cen- ter, it is unreasonable for irregular shape of data distribution. For the intrusion detection model based on rough set and fuzzy support vec- tor machine, dig out the contribution which every condition attributes to decision attribute,propose a new design method of membership function based on weighted comparison between sample and the class center. The model using the method can reduce the geometry de- pendence of the sample set. It can effectively distinguish the sample points and noise points and oufliers. Experimental results show that comparing with support vector machine and traditional fuzzy support vector machine, the fuzzy support vector machine with new mem- bership function can achieve the best classification, and the new method is simple and fast.
同期刊论文项目
同项目期刊论文
期刊信息
位置：