Machine Learning Based Data Access Control in Cloud Computing: A Comprehensive Review

This review article provides a detailed analysis of the application of machine learning (ML) models in data access control within cloud computing environments, focusing on advancements, challenges, and future research directions. The analysis highlights three main aspects: ML-based access control models, traditional access control mechanisms, and performance evaluation metrics. In ML-based models, Deep Learning (47.06%) and Random Forest (20%) were most frequently adopted, showing superior performance in handling large datasets and real-time data processing. Traditional mechanisms like Attribute-Based Access Control (ABAC) were dominant (50%), emphasizing scalability and fine-grained access control. ABAC models were noted for their efficiency in dynamic environments, although they face challenges like computational intensity and poor user experience. Performance evaluation metrics such as accuracy (22%), precision (20%), recall (18%), and F-measure (17%) were predominantly used to assess the effectiveness of ML models. Insights reveal an increasing trend towards deep learning models for their robustness in anomaly detection and adaptability to changing data. The review identifies critical gaps, including vulnerability to adversarial attacks, dataset limitations, and the need for improved computational efficiency. The trends emphasize the importance of integrating ML with traditional methods to enhance the robustness of data access control mechanisms, highlighting the necessity for continuous improvement in algorithm development, dataset quality, and adversarial defenses to secure cloud computing environments effectively.