Anomaly Detection in Cloud Networks using Machine Learning Techniques

This research looks into how machine learning can be used to spot unusual activity in cloud networks, with the goal of making them more secure. It covers eleven key areas of cloud security, focusing on well-known threats like DDoS, Probe, R2L, U2R, and issues around data privacy. The study reviews over thirty different ML methods, with Random Forest (RF) coming up as the most commonly used. Core techniques like SVM, KNN, RF, Naïve Bayes, and Decision Trees were evaluated using thirteen different performance metrics, with special attention given to how accurately they detect threats and how efficiently they train. By analysing twenty datasets—including major ones like KDD and KDD CUP '99—the research shows that ML models can effectively detect and classify abnormal behaviour. It also points to future opportunities for using newer datasets and building adaptive models to keep up with evolving cloud security challenges.