Algorithm Development of Network Intrusion Detection with Adaboost.m1

Abstract

        The objectives of this research were 1) to develop and detect network anomaly with Adaboost.m1 technique and conduct dimension reduction with Gain Ratio and 2) to compare the efficiency of classifying proposed algorithm with Supervised Learning algorithms, This experiment used NSL-KDD database, a network intrusion database. True Positive Rate (TP Rate), False Positive Rate (FP Rate), Precision, Recall, f-Measure, and Accuracy were determined for the performance analysis and comparison.

       The results of this study were as follows: 1) data dimension reduction resulted in important features. When data were classified with Adaboost.m1 technique and decision tree was used as weal learner, it was found that the for highest classification efficiency, the accuracy was 99.79%, 2) When efficiency was compared, the efficiency of proposed algorithm was better than dimension reduction without Adaboost.m1 technique, and classification technique without dimension reduction, when processing time were compared, it was found that proposed algorithm took the highest time, compared to all methods because it required to create a number of models compared to all methods for voting the final answer. For the application with the network anomaly detection, the appropriate method can be selected according to the need.