Anomaly Detection and Response Mechanism of User Sensitive Information Access Control Based on Deep Learning

There are many types of user-sensitive information, and the access modes are complex and changeable. In the actual environment, abnormal access events are usually far less than normal access. This data imbalance will lead to model training difficulties and affect the detection performance. Therefore, the abnormal detection and response mechanism of user-sensitive information access control based on deep learning is proposed. By using the Boltzmann machine (BM) in deep learning, the features of user-sensitive information are selected, and the features of user-sensitive information are fused based on D-S theory. Based on this, the C4.5 decision tree algorithm is used to identify user abnormal access behavior. Combining genetic algorithm and BP neural network, the security response mechanism of user sensitive information transmission link is designed. The experimental results show that the hit rate of access anomaly detection is close to 100%, and the values of F and AUC are close to 1, which indicates that the detection mechanism has high accuracy and reliability in identifying abnormal access. At the same time, this also means that the mechanism does a good job of balancing accuracy and recall, which can effectively reduce false positives and missed positives.