Predictive Cyber Threat Analysis in Cloud Platforms Using Artificial Intelligence and Machine Learning Algorithms

Abstract

In this study, a comprehensive machine learning (ML) framework for threat detection across cloud platforms has been reported. The combinations involved , integrating supervised, unsupervised, and deep learning models. The workflow is presented to consists of data collection, preprocessing, model selection, training, evaluation, and deployment. Quantitative analysis was carried out using datasets from AWS, Azure, and GCP, comprising over 1.2 million log entries. Models were considered and evaluated such as Random Forest (RF), Support Vector Machine (SVM), XGBoost, Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM). The supported the CNN with highest ROC-AUC score (0.94), before LSTM (0.91) and XGBoost (0.87). The predictive framework yielded threat alerts and risk scores approaching an average precision of 92% and recall of 89%. A heatmap evaluation showed the DDoS attacks as the most frequent threat on AWS. However, Insider threats dominated on Azure. The system was deployed with real-time alerting and dashboard visualization, demonstrating scalable performance and actionable insights for cloud security operations.