关键词: 内存特征, BP神经网络, 完整性度量, 组合特征, 内核安全

Abstract: Abstract: In view of the memory data will change after the attack behavior occurs, and the traditional integrity measurement system using benchmark measurement method has the problems of low detection rate and lack of flexibility, a memory combined feature classification method based on multi BP neural network was proposed. The memory data was extracted by measuring object extraction algorithm (MOEA) algorithm, and the model was trained by different BP neural networks.Thena BP neural network was used to collect the data and get the overall safety score of the system. Compared with the traditional benchmark system, the proposed method has higher accuracy and portability. The experimental results show that the multi BP neural network memory combination feature classification method is superior to the traditional single BP neural network classification method in terms of detection accuracy, model complexity and model training time.

Key words: memory feature, back propagation neural network, integrity measure, combination feature, Kernel security