Unlock instant, AI-driven research and patent intelligence for your innovation.

Unbalanced traffic classification method and system based on adversarial generative network traffic enhancement

A traffic classification and network traffic technology, applied in biological neural network models, neural learning methods, other database clustering/classification, etc., can solve problems that are not conducive to obtaining global optimal results

Pending Publication Date: 2022-02-11
INST OF INFORMATION ENG CHINESE ACAD OF SCI
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This is not conducive to obtaining the global optimal result

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Unbalanced traffic classification method and system based on adversarial generative network traffic enhancement
  • Unbalanced traffic classification method and system based on adversarial generative network traffic enhancement
  • Unbalanced traffic classification method and system based on adversarial generative network traffic enhancement

Examples

Experimental program
Comparison scheme
Effect test

example 1

[0073] Example 1 classifies the traffic generated by different user behaviors

[0074] The VPN-NonVPN dataset (aka ISCXVPN2016) is a well-known public dataset of network traffic. It provides 28G real-world traffic captured from ISCX, which mainly includes 7 types of traffic according to different user behaviors and applications. Each type includes two sessions, a regular encrypted traffic session and a VPN protocol-encapsulated traffic session. In the VPN encapsulation traffic session, the number of traffic samples of the majority class is 11.07 times that of the class with the least number of samples. Use TA-GAN to train and test VPN encapsulated traffic, and compare with 7 comparison methods. The results show that TA-GAN is the only method that improves the accuracy, recall, F1 score, and G-mean score of all traffic categories, and the improvement is large. The MAUC (mean area under the PR curve) score was as high as 0.9534.

example 2

[0075] Example 2 classifies VPN encapsulated traffic generated by different user behaviors

[0076] The VPN-NonVPN dataset (aka ISCXVPN2016) is a well-known public dataset of network traffic. It provides 28G real-world traffic captured from ISCX, which mainly includes 7 types of traffic according to different user behaviors and applications. Each type includes two sessions, a regular encrypted traffic session and a VPN protocol-encapsulated traffic session. In regular encrypted sessions, the majority class traffic samples 3.55 times more than the class with the least number of samples. Use TA-GAN to train and test on regular encrypted traffic, and compare with 7 comparison methods. The results show that TA-GAN improves the F1 score of minority traffic by as much as 14.64 percentage points. Moreover, its effect is robust, maintaining the performance of the majority class while improving the performance of the minority class.

[0077]Based on the same inventive concept, anot...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to an unbalanced traffic classification method and system based on adversarial generative network traffic enhancement. According to the method, an empirical optimal network is pre-trained on an original unbalanced traffic data set to serve as an initial state of a classifier; then a generator, a discriminator and the classifier are synchronously trained; the generator oversamples minority class traffic to generate traffic samples, and inputs the traffic samples into the discriminator and the classifier; the discriminator judges whether the input traffic sample is real data or data generated by the generator, and feeds back the data to the generator to help the generator to carry out optimization learning; the classifier classifies the network traffic and feeds a classification result back to the generator, so the generator generates a traffic sample which better conforms to the distribution of the samples of the corresponding category; the output result of the classifier after training is the unbalanced traffic classification result. According to the method, the disadvantage that a general oversampling algorithm is not suitable for traffic data is avoided, and the unbalanced traffic can be effectively classified in a real network environment.

Description

technical field [0001] The invention relates to a method and system for classifying unbalanced traffic based on confrontation generated network traffic enhancement, belonging to the technical field of computer software. Background technique [0002] As an important basic technology for network management and network security, network traffic classification plays an indispensable role in tasks such as service quality optimization, flow-based network billing, and network intrusion detection. With the development of encryption protocols and the improvement of people's requirements for privacy protection, the current network traffic presents a trend of full encryption. The traffic classification technology based on port and rule matching is no longer applicable, and machine learning has become the most mainstream application in traffic classification research. and the most effective technology. However, Internet traffic generally presents a natural uneven distribution. Some hi...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/906G06K9/62G06N3/04G06N3/08
CPCG06F16/906G06N3/08G06N3/047G06N3/048G06N3/045G06F18/2415G06F18/241
Inventor 熊刚李镇郭煜崔明鑫徐安林管洋洋
Owner INST OF INFORMATION ENG CHINESE ACAD OF SCI