Skip to Main Content
Over the past few years, anomaly detection has been an increasing concern with the rapid growth of the network security. Hidden Markov model (HMM) has been applied in various methods in intrusion detection and proved to be a good tool to model normal behaviors of privileged processes, however, one major problem with this approach is that it demands excessive computing resources and costs a long model training time, which makes it inefficient for practical intrusion detection. This paper presents a new method of bringing rough set reduction into HMM to overcome the shortcoming. The proposed approach classifies and simplifies the long observation sequence by virtue of rough set reduction, and the decision conditions obtained in rough set reduction phase could be used in further detection. The experimental results indicate that this method can promote the model training efficiency. Further-more, it is suitable for anomaly detection with high detect rate and low false alarm rate.