The risk evaluation model of network information security based on improved BP neural network
Bin Luo, Yi Liu · 2012
To improve the accuracy and reliability of the risk evaluation of network information security risk, this paper uses rough set attribution reduction to reduce the various factors that affect the network and information security risk, excluding the attributes associated with the decision-making and achieving a typical sample. Besides, in order to train the network, the degree of membership of a typical sample calculated by fuzzy method is as input to neural networks, expert value as the desired output of the network, which can increase the training speed and accuracy. The output of the network can be calculated using the trained network, and network information security risk assessment and decision-making can be achieved based on this output.