A situational-aware method for network security based on residual convolutional neural network
Due to the diversified factors affecting the network security situation,the observed value and forecast value of the network security situation are also constantly changing.This fluctuation leads to the corresponding convergence error of traditional neural networks.The research on network security situational awareness based on residual convolutional neural network.From the perspective of the network itself and the attack state,the influencing factors of the network security situation,and then the weighted average of the overall state parameters of the input neural network to extract the state of the network security situation.The residual loss parameter is introduced to restrict the pooling results of the residual convolution neural network,and the final network security situation value is output.In the test results:the convergence error value shows high stability against different types of network traf-fic and attack means,and is always at a low level,and the maximum convergence error value is only 0.0345.
residual convolutional neural networknetwork security situational awarenessinfluencing factorsquantitative analysisweighted averageresidual loss parameterconvergence error value
NETL
NSTL
万方数据
