Efficient host based intrusion detection system using Partial Decision Tree and Correlation feature selection algorithm

System security has become significant issue in many organizations. The attacks like DoS, U2R, R2L and Probing etc., creating a serious threat to the appropriate operation of internet services as well as in host system. In recent years, intrusion detection system is designed to prevent the intruder in the host as well as in network systems. Existing host based intrusion detection systems detects the intrusion using complete feature set and it is not fast enough to detect the attacks. To overcome this problem, this paper proposes an efficient HIDS - Correlation based Partial Decision Tree Algorithm (CPDT). The proposed CPDT combines Correlation feature selection for selecting features and Partial Decision Tree (PART) for classifying the normal and the abnormal packets. The algorithm is implemented and has been validated within KDD'99 dataset and found to give better results than the existing algorithms. The proposed CPDT model provides the accuracy of 99.9458%.