Application Research on Information System Security Situational Awareness

摘要

Aiming at the problem of high complexity of situational awareness model in which the different test levels, control points and requirements are correlated with each other in the network security level protection test, an improved convolutional neural network security situational awareness method based on one-hot is proposed. Combining the advantages of feature selection of prior selection and posterior selection, the correlation prediction based on multiple different levels and different observation index points is realized. The experimental results show that the proposed method significantly improves the correlation discovery of different parameters between multiple levels, and reduces the classification prediction time. The accuracy of situational awareness is higher than that of deep learning based on feature data.