Viability Assessment about Applicability of Machine Learning in Cloud Anomaly Detection

Abstract

The rapid growth of cloud computing has introduced more opportunities as well as significant challenges in ensuring the reliability and performance of cloud-based systems.Detecting unusual activities such as performance drops, unexpected high resource usage, or security threats, is crucial to avoid disruptions. This work named as “Viability Assessment about Applicability of Machine Learning in Cloud Anomaly Detection (VAAMLCAD)” looks into whether machine learning methods can help to identify and predict anomalies in cloud environment. Different machine learning models in particular Random Forest (RF), Decision Tree (DT), Support Vector Machine (SVM) K-Nearest Neighbor, and K-Means method to see how well aforementioned methods work in identifying anomalies in dynamic cloud environments. A dedicated testbed is constructed with state-of-the-art development frameworks and evaluation tools to measure the performance such as Accuracy Precision, Sensitivity, Specificity, F-Score, and Average Processing Time for the benchmark datasets KDD-Cup and UNSW-NB15. Rank wise discussions based on the performance of the compared methods are vividly elucidated in this work.