Recognition of Access Control Role Based on Convolutional Neural Network

In order to solve the problem that the user-role relationship is difficult to be intelligently allocated in the role engineering of access control in the open environment such as big data, this paper chooses the convolutional neural network to research the access control role recognition. Compared with the traditional user-role assignment method, which is necessary to conduct manual role-granting work for a large number of users through the research and analysis of security experts, the convolutional neural network can be used to omit the artificial role granting process. Based on the existing user's behavior log information in the current information system, this paper transforms the user's behavior characteristics into grayscale images, and proposes an improved convolutional neural network RmNet-7 to automatically extract user behavior characteristics for access control roles recognition. This method has achieved good experimental results in the CERT Insider Threat r2 dataset, and can achieve 80.6% recognition accuracy in the test set.