Data Analytics for Cybersecurity Based on Machine Learning Algorithms

Abstract

Cybersecurity has become one of the key issues in the world. Artificial intelligence (AI), especially machine learning (ML) has been used in the research and applications of cybersecurity such as the research on network intrusion. The database ‘spambase’ is used as an example of network intrusion and cybersecurity in this paper, and data analytics is performed based on radial basis function networks (RBFNs) and multi-layer perceptron (MLP) neural networks, respectively. False positive rate (FPR), false negative rate (FNR), and accuracy are used as evaluation metrics to assess the performance of the two ML methods (RBFNs and MLP neural networks). The data analytics includes the effect of the hidden layer size on the FPR, FNR, and accuracy of RBFNs and the MLP neural networks. The results of RBFNs and MLP neural networks with one hidden layer are compared. Deep learning based on MLP neural networks is also presented.