基于机器学习的入侵检测方法性能评估

IDS performance analysis by applying machine learning approaches

随着网络技术的快速发展,网络恶意攻击方式也逐渐增多,入侵检测系统被开发用于监控和预警企业网络运行状态,保证企业计算机系统的安全。由于现有自适应动态捕获恶意网络数据流的入侵检测系统通常会占用较大的计算机系统资源,为此,文章基于机器学习方法预测网络用户行为和网络数据流分类,针对采用的代表性方法进行对比研究和性能评估,提出能动态适应网络运行状况的轻量级入侵检测系统规则提取技术。

With the rapid development of network technology, there are emerging variously malicious attack over networks.Therefore, kinds of Intrusion Detection System （IDS） are designed and implemented to secure enterprise computer systems by monitoring and predicting the network status.However,most existing dynamic IDS is inclined to be heavy on computer resources when detecting the malicious network traffic.This paper predicts the user behavior and categorized traffic based on representative machine learning approaches,proposes a rule extraction benefit for developing a light and adaptable IDS,and concludes the results from the performances produced by those approaches.