Research on data imbalance in intrusion detection using CGAN

To address the problems of attack category omission and poor generalization ability of traditional Intrusion Detection System (IDS) when processing unbalanced input data, an intrusion detection strategy based on conditional Generative Adversarial Networks (cGAN) is proposed. The cGAN generates attac...

Full description

Saved in:

Bibliographic Details
Published in:	PloS one 2023-10, Vol.18 (10), p.e0291750-e0291750
Main Authors:	Zhao, Guangyu, Liu, Peng, Sun, Ke, Yang, Yang, Lan, Tianyu, Yang, Han
Format:	Article
Language:	eng
Subjects:	Algorithms Analysis Biology and Life Sciences Classification Computer and Information Sciences Data mining Datasets Deep learning Detectors Engineering and Technology Evaluation Feature selection Generative adversarial networks Information management Interpersonal relations Intrusion detection systems Neural networks Performance indices Redundancy Research and Analysis Methods Social Sciences Teaching methods
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	To address the problems of attack category omission and poor generalization ability of traditional Intrusion Detection System (IDS) when processing unbalanced input data, an intrusion detection strategy based on conditional Generative Adversarial Networks (cGAN) is proposed. The cGAN generates attack samples that approximately obey the distribution pattern of input data and are randomly distributed within a certain bounded interval, which can avoid the redundancy caused by mechanical data widening. The experimental results show that the strategy has better performance indexes and stronger generalization ability in overall performance, which can solve insufficient classification performance and detection omission caused by unbalanced distribution of data categories and quantities.
ISSN:	1932-6203 1932-6203