A KNN-ACO approach for intrusion detection using KDDCUP'99 dataset

Feature reduction in data mining from the large dataset can proficiently develop and enhance the performance of the whole system. The reduction of feature is chosen using entropy and information gain. Consequently, that reduced feature subset is passed to the KNN classifier to expose the normal and abnormal classes of attack. In this work, intrusion detection using data mining algorithm is discussed. The intrusion detection system supervises the activity performed by the users and if any type of misconduct activity is detected then it alert to the administrator to avert it. Here, the exercise is done on KNN classifier with ACO (ant colony optimization) method to expose the intruders and make a use of KDDCUP dataset to categorize different class of attack. The simulation analysis of proposed system is done using accuracy and false alarm rate (FAR) performance parameter. Our proposed system generates more accurate results than the existing method.