Journal of Artificial Intelligence, Virtual Reality, and Human-Centered Computing
Advancing Insider Threat Detection: A Novel Framework for Enhanced Security in Cloud Computing Environments
Abstract
Segun Kazeem Fatoki, Olalekan Akinbosoye Okewale, Mayowa Oyedepo Oyediran, Olufemi S Ojo and Olugbenga Ayomide Madamidola
Insider threats are major security issues in cloud computing where legitimate users with privileged access misuse their credentials to attack data, systems, or services. Conventional intrusion detection systems are ineffective for insider threat detection in cloud computing since they are based on predefined rules or signatures that are not well-suited for cloud dynamics. In this study, a novel insider threat detection model for cloud computing is proposed by combining the capabilities of convolutional neural networks (CNN) and gated recurrent units (GRU) to extract both spatial and temporal information from the Community Emergency Response Team (CERT) insider threat dataset containing 30,000 samples from Carnegie Mellon University. The proposed CNN-GRU hybrid network performs better with an accuracy of 99.8%, a sensitivity of 99.7%, and a lowest false negative rate of 0.002%, outperforming the accuracy of 96.98% achieved by CNN and the accuracy of 95.91% achieved by LSTM.