Validation of Decisions of a Multilayer Perceptron Learning Algorithm for the Identification of Net Attacks with the Aid of Bayesian Classifiers

An intrusion detection system (IDS) is a software application that monitors the network for potential malicious attacks against a single computer or a computer network. A multilayer perceptron (MLP) learning algorithm is used detect such attacks and identifies the kind of attack like WebAttack, DoS or BruteForce. A multilayer perceptron (MLP) is a class of feedforward artificial neural network (ANN), which consists of at least three layers of nodes: an input layer, a hidden layer and an output layer. Since ANNs belong to the so called black box algorithms, it is useful to validate its results. In this paper a method is presented to validate the decisions of the MLP algorithm concerning the type of net attack with the help of Bayesian Classifiers. Particularly the Naïve Bayesian Classifier and the Tree Augmented Naïve (TAN) Bayesian Classifier are used for this task. It will be shown that these classifiers are capable to satisfactorily validate the decisions of the MLP algorithm. This will be accomplished with aid of real datasets from the Canadian Institute for Cybersecurity along with appropriate metrics to evaluate Machine Learning algorithms.