Intrusion detection using rough set classification

摘要: Recently machine learning-based intrusion detection approaches have been subjected to extensive researches because they can detect both misuse and anomaly. In this paper, rough set classification (RSC), a modern learning algorithm, is used rank the features extracted for detecting intrusions generate models. Feature ranking very critical step when building model. RSC performs feature before generating rules, converts minimal hitting problem addressed by using genetic algorithm (GA). This done in classical Support Vector Machine (SVM) executing many iterations, each of which removes one useless feature. Compared with those methods, our method avoid iterations. addition, hybrid proposed increase convergence speed decrease training time RSC. The models generated take form “IF-THEN” advantage explication. Tests comparison SVM on DARPA benchmark data showed that Probe DoS attacks yielded highly accurate results (greater than 99% accuracy testing set).