Hybrid Algorithm of Differential Evolution - Support Vector Machine (DE-SVM) for Network Intrusion Detection System

Along with the increasing amount of important data stored on the server computer, there is a greater need to secure the network connected to the server. Several researchers have proposed techniques that utilize artificial intelligence and machine learning. In this study, we evaluate the intrusion detection capability of a support vector machine (SVM), which is optimized using the differential evolution (DE) algorithm. We used an SVM model without parameter tuning and a hybrid model, PSO-SVM, as comparators. In this study, a deep learning model, that is, a deep convolutional neural network (DCNN), was also used as a comparator. All models were trained and evaluated using the training and test sets from the NSL-KDD dataset. Each model was evaluated using a classification accuracy metric for the test set. From the experimental results, we conclude that the hybrid DE-SVM model yields better results than the SVM model.