Anomaly Detection in Cloud System using Novel Aspect of SMOTE Sampling and Machine Learning Classifiers

Misra, Gauri (2023) Anomaly Detection in Cloud System using Novel Aspect of SMOTE Sampling and Machine Learning Classifiers. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

The main problem of class imbalance in machine learning (ML) models is addressed in this study, which presents a novel approach to improve anomaly detection in cloud systems. To increase the detection accuracy of uncommon anomalies which are usually underrepresented in cloud datasets—the main contribution is the combination of powerful machine learning classifiers with the Synthetic Minority Oversampling Technique (SMOTE). Various models, including Deep Neural Networks (DNNs), Recurrent Neural Networks (RNNs), Gradient Boosting Machine (GBM) and Random Forest (RF) were thoroughly evaluated as part of the research. Findings demonstrated that Deep Neural Network and Gradient Boosting Machines were capable to identify outliers with pinpoint accuracy. In terms of cloud security, the significance of this research is the comprehensive methodology of employing the SMOTE sampling techniques. The proposed research assesses the various machine learning models that will detect anomalies in cloud computing with higher accuracy. New avenues for research include developing and testing data balancing algorithms with more sophisticated features, and study hybrid models that combine the best features of different approaches. The essential quality in the ever-evolving world of cyber threats, these endeavours may produce innovative and highly adaptable security solutions for cloud computing. The four classification models that have been created for the detection of the anomaly in cloud environment, the Random Forest and the Deep Neural Network models achieved reliable accuracy of 0.98 and 0.99 respectively. However, the RNN classifier model achieved a poor accuracy of prediction which is 0.14. An excellent result of 100% accuracy is achieved by the Gradient Boosting Model (GBM). All the performance parameter values of GBM Model are equal to 1.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Mijumbi, Rashid UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	09 Apr 2025 11:17
Last Modified:	09 Apr 2025 11:17
URI:	https://norma.ncirl.ie/id/eprint/7391

Actions (login required)

View Item