A Hybrid Two-Step-Model Based on Cuckoo-Search and Grey Wolf Optimiser Ensemble-Based Classifier Approach for Network Intrusion Detection

The new generation of Internet threats brought higher risks over network connectivity. This research suggests a layered feature selection Intrusion Detection System with an ensemble learning algorithms to enhance the decision about possible intrusion attacks. Cuckoo Search and Grey Wolf Optimiser approaches have been proposed to inherit their advantages to select the most distinctive features. The proposed classification model is combined with five classifiers, Fuzzy Unordered Rule Induction Algorithm, Voted Perceptron, Simple Logistic, Forest by Penalizing Attributes and C4.5 machine learning algorithms, to form a single voting-based ensemble classifier. The result indicated that the performance of the proposed ensemble based on the maximum of probabilities rule is better than the individual classifiers.