A Machine Learning Based Hybrid Encryption System to Prevent Cloud Data Breach

Abstract

As cloud computing becomes increasingly central to data storage and processing, the need for robust security mechanisms to protect sensitive information during cloud uploads is more critical than ever. This research presents a novel hybrid security framework that combines symmetric (AES) and asymmetric (RSA) encryption techniques with a machine learning-based Intrusion Detection System (IDS) to secure data transmissions in cloud environments. The proposed model addresses key challenges such as insider threats, data breaches, and insecure APIs by employing a two-tier approach: encrypting data for confidentiality and using ML-driven IDS to detect malicious patterns in real time. The system was evaluated using the CICIDS2017 dataset and implemented in a simulated cloud setting. Performance analysis demonstrated that the hybrid model outperforms standalone encryption or IDS systems in terms of detection accuracy, encryption speed, resource efficiency, and resilience against various attack vectors. The results support the model’s suitability for secure, scalable, and intelligent cloud data management, offering a future-proof solution adaptable to evolving cyber threats.