Towards Proactive Cloud Security: A Survey on ML and Deep Learning-Based Classification Techniques on Network Intrusion Systems

Abstract

Cloud computing's scalability, flexibility, and on-demand resource availability have revolutionized contemporary IT architecture. However, complicated security and privacy issues are brought up by its dynamic, distributed, and multi-tenant nature. When it comes to dealing with these changing threats, conventional rule-based intrusion detection systems (IDS) frequently fail. With an emphasis on proactive and adaptable security measures, this study offers a thorough review of IDS based on machine learning (ML) and deep learning (DL) for cloud settings. The review examines the effectiveness of supervised and unsupervised ML models, as well as advanced DL architectures, in real-time threat detection and anomaly prediction. Behavior-based analysis, feature selection techniques, and predictive models such as Facebook Prophet are highlighted for their role in improving detection accuracy. Key datasets, evaluation metrics, and deployment tools are also discussed to offer practical insights into the implementation of intelligent IDS solutions. Emphasizing a shift from reactive defense to predictive, self-learning strategies, this survey outlines the potential of AI-driven IDS to enhance scalability, accuracy, and resilience in cloud security. The findings aim to support future research and development of next-generation intrusion detection frameworks for secure cloud computing.