Enhancing Intrusion Detection Systems Accuracy Using Machine Learning

Following the digital revolution, massive amounts of data have been produced over time through a variety of networks. The networks have made data analysis exceedingly challenging by identifying attacks using the appropriate approaches. Although intrusion detection systems (IDSs) protect resources from threats, they still struggle with increasing detection accuracy, cutting down on false alarm rates, and finding unidentified threats. Making machines intelligent is the aim of artificial intelligence, and knowledge acquisition is made possible via machine learning. Making a computer self-learner is the goal of machine learning (ML), a sub-field of artificial intelligence. This capacity for self-learning will be useful in tackling many challenging issues. Utilizing a machine learning intrusion detection system can increase its effectiveness and ability to recognize new attack patterns through self-learning or knowledge acquisition. IDS are the first line of defense because they gather data or knowledge from a network and analyze it to identify components that are breaking network and computer security laws. In this paper, some machine learning algorithms are going to be implemented onto an IDS to measure their detection accuracy. The K-Fold cross validation method is going to be performed in order to increase the detection rate. As well as run the algorithms multiple times in order to find out which features hold more importance when compared with other features. The machine learning algorithms K-Nearest Neighbor (KNN), Decision tree, and Random Forest are going to be implemented. RF with 100 trees outperformed the other algorithms with an accuracy of 92.65%.