A Comparative Study of Machine Learning Methods for Intrusion Detection

In this work, we applied 8 machine learning (ML) techniques to detect intrusions, namely, neural networks, kNN, SVM, random forest, trees, AdaBoost, naive Bayes, and stochastic gradient descent SGD. Using the NSL-KDD data set, these ML techniques were trained and tested to correctly classify the network and operating system records of this dataset into one of 24 possible attacks. The performances of these ML methods were analyzed and compared, with the random forest method performing at the top. To the best of our knowledge, this is the first work on investigating more than 4 ML classifiers on this data set in one single work and using the same set of tools.