An optimal feature subset selection technique to improve accounting information security for intrusion detection systems

The goal of an intrusion detection system (IDS) is to secure data on the network by monitoring the flow of traffic and identifying malicious users. Since there is enormous data available on the network, an optimal selection of features is necessary to get rid of irrelevant and redundant data. Therefore, the given paper introduces a novel technique by making use of a modified firefly optimization algorithm (MFOA), followed by the use of an autoencoder (AE), to improve accounting information security in IDS’s. MFOA acts as an optimal feature selection method that works on the concept of random firefly generation, thereby finding the best feature subset. Now, the reduced set of features is fed as an input to AE for the detection of intrusions and anomalies in the network. The AE comprises an encoder–decoder for generating the reconstructed input dataset. The NSS-KDL dataset is taken into consideration, and the performance of the proposed approach is validated based on evaluation metrics such as detection rate (DR) and accuracy (ACC). The results show that the proposed approach attains higher DR and ACC than the existing machine learning (ML) and deep learning (DL) methods used in the given study.