An Intelligent Federated Learning Framework for PrivacyAware Threat Detection in Secure Cloud Computing Environments

Abstract

The widespread adoption of cloud computing has enabled organizations to deliver scalable digital services and manage distributed computing resources efficiently. However, the continuous exchange of sensitive information across cloud infrastructures has intensified concerns regarding data confidentiality, user privacy, and cyber resilience. Conventional machine learningbased security solutions often rely on centralized data collection, which increases the risk of unauthorized data exposure, regulatory non-compliance, and privacy breaches. These limitations highlight the need for collaborative security mechanisms that can learn from distributed data while preserving its confidentiality.This study proposes an intelligent federated learning framework for privacy-aware cloud security that enables multiple cloud participants to jointly develop an effective threat detection model without sharing their raw datasets. The proposed framework performs local model training at each participating node and securely exchanges only encrypted model parameters for global model aggregation. This decentralized learning strategy minimizes data transmission, protects sensitive information from external disclosure, and strengthens compliance with modern privacy requirements. Machine learning techniques are employed to recognize malicious behaviors, abnormal access patterns, and emerging cyber threats by learning from diverse security events generated across distributed cloud environments. The framework further incorporates secure aggregation and iterative model optimization to improve learning efficiency while maintaining data integrity and model robustness.The effectiveness of the proposed approach is assessed using standard performance metrics, including accuracy, precision, recall, F1-score, and communication efficiency, to evaluate both security performance and collaborative learning capability. Experimental analysis demonstrates that the federated learning framework achieves reliable threat detection while significantly reducing privacy risks associated with centralized data processing. By combining decentralized artificial intelligence, privacy-preserving model training, and collaborative threat intelligence, the proposed framework provides a scalable and secure solution for protecting modern cloud infrastructures against evolving cyber threats without compromising the confidentiality of organizational data.