基于随机森林的网络攻击检测方法

摘要

网络攻击检测是网络领域的一个重要的应用，目前在这领域内的检测方法有很多，但是已有的检测机制存在着错误率高以及无法处理数据不平衡等问题。通过分析网络攻击数据，设计了基于随机森林的网络入侵检测算法，并把这个算法用于网络连接信息数据的检测和异常发现。通过对CUP99数据的测试集进行试验，基于随机森林的算法能够提高识别效率，有效的解决数据不平衡带来的问题，具有很好的分类效果。%Network intrusion detection is one of the important application in network area. At present, there are various detection approaches in this area. However, some problems are found in the existing algorithms, including high error rate and failure processing of data imbalance. After analyzing the network intrusion data, we design an intrusion detection algorithm based on random forests and apply it to detection of network connection information data and anomaly find. The experiment on CUP99 dataset proves that this algorithm improves identification efficiency, effec-tively solves the problem of data imbalance, and shows a better classification effect.