Artificial Intelligence for Cyber Threat Detection in Cloud Computing: A Hybrid Random Forest and LSTM Approach

Abstract

Cloud computing has become integral to modern enterprise infrastructure, yet its open and dynamic nature exposes it to sophisticated cyber threats. Traditional intrusion detection systems (IDS) lack adaptability and temporal awareness, limiting their effectiveness against evolving attacks. This paper proposes a hybrid AI/ML-powered intrusion detection framework that leverages the classification strength of Random Forest (RF) and the sequential learning capability of Long Short-Term Memory (LSTM) networks. An adaptive feedback mechanism continuously refines the system by learning from false positives and emerging threats. Experimental results on benchmark cloud intrusion datasets demonstrate that the proposed model achieves superior detection accuracy, reduced false positive rates, and improved adaptability compared to classical methods.