Author(s)

Tuqian Zhang

Corresponding Author

Tuqian Zhang

Abstract

With the rapid development of information technology, the neural network model has developed rapidly in recent years. At the same time, the high latitude and non-linear characteristics of computer network data make the network intrusion detection work difficult to break through, and the network security problem has become the focus of problems in all walks of life. In this context, this paper deeply analyzes the characteristics of deep neural network and network intrusion detection, and constructs an intrusion detection model based on deep neural network by using functions such as ReLU activation function and cross entropy loss. The experimental results show that the influence of network parameter selection on the experimental results is small, and the experimental results of the spindle-shaped network structure are better than the pyramidal network structure.

Keywords

Deep neural network; Computer; Network intrusion; Application