Enhancing Security in Cloud Computing With Anomaly Detection Using Random Forest

Enhancing security in cloud computing represents a crucial initiative that leverages cutting-edge techniques and equipment to identify and thwart malicious activities. An integral approach in this regard is anomaly detection, which entails vigilant monitoring of network traffic, user behavior, and system resource utilization to pinpoint unusual patterns indicative of potential threats jeopardizing confidential data. This abstract explores the application of Random Forest, a machine learning algorithm tailored for anomaly detection in cloud computing, with a specific focus on addressing threats like Worms and Denial of Service. Renowned for its effectiveness, the Random Forest algorithm exhibits robust capabilities in near real-time anomaly detection. This involves constructing multiple decision trees during the training phase and determining the class mode (for classification) or mean prediction (for regression) based on the collective output of these individual trees. Noteworthy advantages of Random Forest include its adept handling of large datasets, ability to manage missing data, and provision of a measure of feature importance. To assess the efficacy of Random Forest for anomaly detection in cloud computing, we utilized a dataset from Kaggle, comprising raw network packets from the UNSW-NB 15 dataset generated using the IXIA Perfect Storm tool within the Cyber Range Lab of the Australian Centre for Cyber Security (ACCS). Performance comparisons were made with other machine learning algorithms, considering accuracy and false alarm rate. Experimental results demonstrate that Random Forest outperforms alternative algorithms, particularly in the context of the k-cross validation method. Random Forest emerges as a robust machine learning algorithm capable of enhancing security in cloud computing through effective identification of anomalous behavior. Its proficiency in handling large datasets, addressing missing data, and providing insights into feature importance positions it as an optimal method for detecting anomalies in cloud computing. This research aims to encourage further exploration of Random Forest for anomaly detection in the cloud computing domain.