基于改进决策树的入侵检测算法

本文对经典的基于信息增益的决策树算法进行改进,提出一种基于决策树与属性相关性相结合的入侵检测算法。该算法同时结合综合策略的剪枝算法以避免过度拟合对检测结果的影响。实验结果证明,本算法不仅在面对已知攻击时能够做出良好的判断,而且在面对未知攻击时仍然具有一定的检测能力,具有良好的性能和可用性。

Improved decision tree based intrusion detection algorithm

This paper improves classical information gain based decision tree algorithm and then presents a decision tree and property relation combined intrusion detection algorithm.The algorithm combines the pruning algorithm of integrated strategy to avoid the impact of excessive fitting on test results.Experimental results show that the algorithm can not only make better judgments when confronting known attacks but also still have certain detection capability when facing unknown attacks.The algorithm thus has better performance and availability.