Authors: Karupothu Bharath, Assistant Professor T Swapna Sridevi
Abstract: Cloud computing has become a fundamental part of modern digital infrastructure, enabling organizations to deploy applications, store data, and deliver services across highly distributed environments. However, the dynamic nature of cloud systems also creates complex security challenges, including abnormal network activity, unauthorized access, misconfigurations, and emerging vulnerabilities. Traditional security mechanisms often depend on predefined rules and known attack patterns, making them less effective when dealing with previously unseen or continuously evolving threats. This paper presents a machine learning-based approach for strengthening cloud security by combining anomaly detection and vulnerability classification. The proposed framework uses an Autoencoder to identify unusual patterns in cloud-related data and XGBoost to classify detected security events and potential vulnerabilities. The system processes security-relevant information such as network traffic, system logs, user activity, and historical vulnerability records through data preprocessing and feature extraction stages before applying the machine learning models. By combining unsupervised anomaly detection with supervised classification, the framework is designed to provide a more adaptive approach to identifying both familiar and potentially unknown security threats. The proposed solution is implemented as a web-based security framework using Python and Django, providing mechanisms for data management, model evaluation, prediction, and security monitoring. It is intended to reduce dependence on manual analysis while supporting faster identification of suspicious activities and potential vulnerabilities. The framework also provides a foundation for continuous model improvement as new security data becomes available. Overall, the proposed approach demonstrates how machine learning can be integrated into cloud security workflows to support proactive vulnerability detection and improve the adaptability of security monitoring in distributed environments.
