基于条件信息熵的网络攻击特征选择技术

Technology of Attack Feature Selection Based on Conditional Information Entropy

用知识的条件信息熵定义了特征的相对重要性,建立一种基于条件信息熵的网络攻击特征选择算法,选出的特征属性不仅可以大大减少数据在存储、分析以及各组件共享中的代价,还能够降低构建神经网络系统的复杂性,简化训练集,减少检测时间,保持并提高入侵分类的准确性.

This paper defined the importance of attack features using conditional information entropy of knowledge, and presented an algorithm of attack features selection based on conditional information entropy. The experiment results show that it is not only effective to reduce the cost of memory and analysis, but also to simplify neural network structure and train set by using the selected features. The results also show that it can increase the detection speed and without sacrificing the detection correctness.