Ensuring Anomaly-Aware Security Model for Dynamic Cloud Environment using Transfer Learning

Cloud concepts such as resource sharing, outsourcing, and multi-tenancy create significant challenges to the security community. Also, trusted third party and web technologies based cloud service provisioning arises new security threats in the cloud environment. Cloud security research still faces the shortcomings in improving the detection accuracy and detecting the new or unknown attacks in the cloud. Machine learning techniques play a significant role in automatically discovering the potential difference between legitimate and malicious data with high accuracy. Hence, it is essential to develop an intelligent security mechanism to learn, adapt, and detect the attacks or anomalies for the distributed and dynamic cloud environment. This work is about security solutions developed by a new mechanism called transfer learning techniques for the cloud environment. The transfer learning model leverages the detection of different types of known and unknown attacks by the utilization of the knowledge from the source domain. Rather learning about the attack from scratch, transfer learning focuses on the transfer of knowledge from source trained attacks to target attacks. This work gives a scope of detecting and solving new attacks on target to be only trained and fix it on the source and maintains qualitative performance.