Reliable Anomaly Detection in Heterogeneous Cloud Data Using Chi-Square Feature Selection and RF Classification
Dharmesh Dhabliya, D. Anandhasilambarasan, Abhiraj Malhotra, S Srisathirapathy, Mohit Gupta, Subbulakshmi Ganesan · 2024
The scalability and dependability offered by cloud infrastructure are important to the modernization of many services and businesses. Data storage and retrieval procedures like this have the potential to revolutionize many areas, including the human life cycle, business, education, and more. The term "intrusion detection" refers to the steps used to find instances of unauthorized access to cloud infrastructure that breaches established security protocols. Unauthorized acts may be more easily detected if the detection procedure is based on the assumptions about non-legitimate users' behaviours, which are more anomalous than usual. Because they are caught and dealt with before any harm comes to the system, these abnormalities are considered to contribute to its dependability. The machine learning algorithms work by analysing past data to make predictions about the future based on what the system has learned, which helps to identify inappropriate behavior. The suggested Random Forest (RF) classifier, which incorporates the ChiSquare algorithm framework, has been used for anomaly detection. This study employs the node selection method for efficient node selection and the Chi-Square technique for identifying the important characteristics for detection. This trustworthy source utilizes the cloud computing environment to identify abnormalities using this anomaly detection approach. Slave nodes do analytics in the same cloud. The workload is significantly decreased even though the node selection technique is used to determine the appropriate number of nodes.