MACHINE LEARNING APPROACHES AND CHALLENGES IN ANOMALY DETECTION SYSTEMS WITHIN CLOUD ENVIRONMENT: A REVIEW

Abstract

Data sharing, Consumption based pricing, Definite services, Fast adaptability ,joint resource access are some of the key areas of Cloud computing popularity. A abundant users can use the stashed framework of cloud and more number of transactions are generated. Due to this cloud security has significant importance with the help of cloud security users can design policies and control accesses to protect the cloud environment from different kinds of threats. A data which is generated during transaction which shows abnormal behaviour, pattern or operations is called as anomaly. The anomalous data can create the troubles like security threats, system malfunction, Efficiency decrease can be identified in the different ways like unauthorized access, network flow pattern, and unpredicted system performance, inappropriate use of resources or anomalous system activity. Finding anomalous activity is crucial because of the vast amount of data that is shared in cloud environments during different transactions. Strengthen security, offer protection against valuable infrastructure, maintain uninterrupted system monitoring, and improve cloud anomaly detection performance in order to guarantee seamless cloud operation. By employing anomaly detection methods, cloud users and providers can bolster the security of cloud infrastructures, thereby reducing the risk of vulnerabilities and attacks. .Three important methodological domains are highlighted by our analysis: statistical techniques, deep learning, and machine learning. and outlines how each model is specifically applied for anomaly detection. Additionally, We outline the typical application areas of anomaly detection within cloud computing systems, as well as the public datasets frequently utilized for assessment. Finally we. make recommendations for future study directions and talk about the consequences of our findings.