Network anomaly detection system and detection method based on a neural network

Abstract

The invention provides a network anomaly detection system and detection method based on a neural network. The detection system comprises a coding processing module, a data normalization module and a feature selection module, an accuracy rate module and an observer operation characteristic curve drawing module. The detection method comprises the following steps: firstly, processing discrete characteristics in a KDCUP 99 data set into a numeric type by adopting one-hot coding; then, carrying out feature processing by adopting Min-Max; secondly, carrying out dimension reduction processing and inputting the processed data into an MLPClassifier multi-layer perceptron classifier to obtain a prediction result; and finally, inputting the prediction result into an observer operation characteristiccurve drawing module to draw an ROC curve. A multi-layer perceptron neural network is adopted, over-fitting is prevented through an L2 regularization method, hidden nodes are adjusted, meanwhile, a cross validation method is adopted for continuous training and debugging, compared with a KNN method and an SVM method, the operation time and the accuracy are superior to those of the two methods, andthe superiority of the method is verified.

Description

technical field [0001] The technology relates to the field of neural networks, in particular to a network anomaly detection system and detection method based on neural networks. Background technique [0002] Due to the wide application of computer networks, it is inevitable to detect network attacks and protect information security. The extensive use of computer systems has caused many threats, and the wide spread of networks has led to various types of attacks, such as zero-day vulnerability attacks. The development of computer networks has greatly exacerbated computer security issues, especially the network security of today's network environment and advanced computing equipment. Although the Internet protocol suite was not designed for security issues, network administrators must now deal with individuals and individuals with malicious intentions. Mass intrusions of large botnets. According to the Symantec Network Security Threat Report, there were more than 3 billion ma...

AI Technical Summary

Problems solved by technology

In addition, the specialized skills needed for cybercrime have declined as cracking tools have become more accessible

Images