运用C4.5算法优化入侵检测系统的研究

研究了以决策树结构取代原有入侵检测系统采用的链表结构,在于改进入侵规则的存储和模式匹配,提高检测速度.用C4.5算法为决策树每个规则集节点动态选择最显著的特征,进行并行测试,实现性能优化.为了减少冗余比较和无效匹配,运用了一种基于C4.5决策树裁剪算法,以规则信息量作为判断标准.研究结果表明,此方法可以提高最终模型的预测精度,并能克服噪音数据,有效地提高了模式匹配速度和规则的访问速度.

Study of C4.5 algorithm based intrusion detection system

Studied a decision tree to replace the existing structure of the chain structure intrusion detection system.It aimed at improving the storage and pattern matching to improve detection rate.Meanwhile,the use of C4.5. Decision tree algorithm for rule set each node dynamic selection of the most notable features parallel testing and achieve optimal performance. To minimize redundant and ineffective compared match, a cut on C4.5 decision tree algorithm,Information as to the criteria for judging rules. The experimental results show that this method can improve the forecast accuracy of the final model, and good overcome noise,effectively improved the speed and pattern matching rules access to the speed.