Intrusion detection method based on partial deep learning theory

Abstract

The invention relates to an intrusion detection method for detecting network data by using a partial deep learning theory. The intrusion detection method comprises the following steps: preparing a data set: selecting a preprocessed training data set and a preprocessed test data set; establishing a CNN and training the CNN, and training the CNN by using the preprocessed training data set; establishing a decision tree DT, and training the DT by using the preprocessed training data set in the first step so as to realize the first binary classification of the test data; performing PCA dimension reduction processing: introducing PCA to perform dimension reduction processing on test data of which the result is normal data after DT classification; and carrying out secondary dichotomy on the testdata of the normal data subjected to PCA dimension reduction processing by utilizing the trained DNN.

Description

technical field [0001] The invention belongs to the field of deep learning and network security, in particular to an intrusion detection method for detecting network data by using part of deep learning theory. Background technique [0002] With the continuous progress of the times, the interconnected network has made people's lives more convenient, and all that is needed for shopping and travel is a mobile phone that can be connected to the Internet. The price of this convenience is that almost everyone’s identity information, social relationship, and personal property are firmly bound to the network. When using the network as a bridge for social interaction, one of the issues that needs to be considered is the "bridge" The stability of the network, that is, the security of the network. Communication systems and network entrances are always facing network attacks from the outside or even within the system, and unlike the single attack in the immature period of the network, ...

AI Technical Summary

Problems solved by technology

Liu Jinghao [3] proposed an intrusion detection model ICA-DNN based on independent component analysis ICA and deep neural network DNN. The combination of intrusion detection and deep learning methods endows the model with better feature learning ability and more accurate classification ability, but The prediction time of the model has not been specifically evaluated, and the real-time performance of the model is poor

[0003] Considering the incompatibility between detection accuracy and detection speed in the above method, the present invention proposes an intrusion detection method combining multiple deep learning theories, which can achieve a relatively high detection rate and have faster detection Speed, effectively solving the real-time problem of intrusion detection

Images