Intrusion Detection Using Supervised Learning with Feature Set Reduction
Intrusion detection systems intend to recognize attacks with a low false positive rate and high detection rate. Many feature selection methods introduced to eliminate redundant and irrelevant features, because raw features may abbreviate accuracy or robustness of classification. In this paper, the authors are proposing the information gain technique for the selection of the features. A feature with the highest information gain is the criteria for the selection of the features. They reduced the features of the data set than run the algorithm. Result show that drastically decreased in learning time of the algorithm without compromising the accuracy which is desirable for good IDS. They analyze two learning algorithms (NB and BayesNet) for the task of detecting intrusions and compare their relative performances.