Analysing Network Intrusion Data

Developmening systems applying machine learning and data mining techniques is one of the approaches to combating network intrusion.Many IDS Intrusion Detection Systems)suffer from a high rate of false alarms and missed intrusions.Tha challenge is to be able to improve the intrusion detection rate at a reduced false positive rate. To counter imbalance in data, a combination of oversampling (synthetic generation using SMOTE) and undersampling techniques had been employed. Two classifier models have been evaluated, Random Forest and Support Vector Machines. Data being used is the KDD Cup 99 data. The results of the experiment shows reasonable improvement in evaluating minority classes through oversampling and under sampling techniques.