入侵检测与聚类分析

传统的入侵检测方法在面对网络结构升级和未知攻击时 ,缺乏必要的扩展性和自适应能力 ,而基于机器学习的检测算法首先需要训练数据集进行训练 ,然后建立检测模型并通过测试数据集中入侵行为的检测结果来验证 ,此类方法由于获取类标识数据的困难性及其信息表达的局限性 ,降低了对未知攻击的检则能力。本文提出利用遗传聚类进行入侵检测算法IDUGC(IntrusionDetectionUsingGeneticClustering)。实验结果表明 ,此算法在未知入侵检测方面是可行的、有效的 ,并具有良好的可扩展性

Intrusion Detection And Clustering Analysis

Traditional intrusion detection methods lack extensibility and adaptability in the face of upgraded network architectures and unknown attacks. Meanwhile, detection algorithms based on machine learning need train data sets to train firstly, then the detection module is set up and validated by the detection result to intrusions in test data sets. Because of difficulties in obtaining labeled data and limitations to the knowledge expression to labeled data sets, the ability to detect unknown attacks is finally degraded. In this paper, Intrusion Detection Using Genetic Clustering (IDUGC) is proposed. By means of simulated experiments, this algorithm is proved feasible, efficient and extensible for unknown intrusion detection.