Enhancing Cloud Infrastructure Security Through AI-Powered Big Data Anomaly Detection
Sunil Jacob Enokkaren, Varun Bitkuri, Raghuvaran Kendyala, Jagan Kurma, Jaya Vardhani Mamidala, Avinash Attipalli · International Journal of Emerging Research in Engineering and Technology · 2021
Numerous resources and computer capabilities are made available over the Internet via cloud computing. Because of its appealing characteristics, cloud systems draw a lot of users. Cloud systems may still have serious security problems despite this. Accordingly, it’s crucial to develop a system capable of detecting abnormalities in cloud environments, allowing for the high detection rate of both insider and outsider assaults. The suggested approach makes use of cutting-edge ML models. XGBoost and Multi-Layer Perceptron (MLP) combined with the necessary preprocessing techniques, i.e., feature selection and SMOTE-based class balancing, are accurate and resilient to identify anomalies in the context of a complex cloud environment. The XGBoost model performed better than other classifiers with 97.5 percent accuracy and 1.00 ROC-AUC. The Multi-Load Pump model also showed excellent results with 96.20 percent accuracy and 0.99 ROC-AUC. The superiority of the suggested models in comparison with conventional methods such as Naive Bayes (NB) and Random Forest (RF) is proved with the assistance of comparative analysis. In general, AI and big data analytics have transformed into a scalable, dependable, and proactive cloud automation framework to secure cloud environments against even advanced cyber threats