An Intrusion Detection System for Identifying Simultaneous Attacks using Multi-Task Learning and Deep Learning
Saleh Albelwi
Abstract
Though both machine learning and deep learning have made great progress in developing intrusion detection systems (IDSs), two critical problems remain. The first is the ever-evolving nature of malicious activity. Data flows and packets can contain several types of attacks simultaneously, while machine learning and deep learning algorithms can only learn a single task at a time. Second, the publicly available datasets are created to evaluate only one type of attack. In order to address these issues, this paper proposes a Multi-Task Learning (MTL) model for an IDS based on deep neural networks (DNNs), which has the ability to detect several types of attacks simultaneously. To do this, we concatenated each corresponding sample in both the UNSW-NB15 and CICIDS2017 datasets into one feature vector. This guarantees that a portion of the data flows, in both the training and testing sets, contains two threats at one time. The experimental results demonstrate that the proposed system maximizes performance.