A Cloud User Behavior Authentication Model Based on Multi-label Hyper-network

Abstract

With the advent of the Big Data era, user information security is particularly important. How to build trust between users and the cloud is an important issue. In response to this problem, this paper proposes a cloud user behaviour authentication model based on multi-label hyper-network, which implements the fine-grained division of user behaviour and improves the accuracy of anomaly detection. This method trains the user’s normal behaviour database into a hyper-network, adds the current user behaviour as an instance to the hyper-network for classification. If a label is successfully found in this classification, it is identified as a normal user. Otherwise, the model updates the weight of the hyper-network, replaces the super edge, and looks for the label again. If the label is found, it is identified as a risk user, otherwise it is identified as malicious user. The simulation results show that there is a significant improvement in the accuracy of classification. Applying the method in this paper to the detection of user behavior can effectively improve the detection rate of user behavior, realize fine-grained analysis of user behavior, and improve the processing ability of user behavior.