Title: A novel LightGBM-based industrial internet intrusion detection method

Authors: Zhiqiang Lv

Addresses: Department of Information Construction, Jilin Institute of Chemical Technology, Jilin City, Jilin, China

Abstract: This paper proposes an Active Learning-based Intrusion Detection System. The system introduces expert annotation into the intrusion detection process, and combines the active learning query strategy with LightGBM to solve the problem of low accuracy of the intrusion detection system when the training samples are scarce. First, the process of data pre-processing is applied. Features are extracted from the traffic, and the borderline SMOTE method is introduced to improve the samples distribution. Then, the LightGBM algorithm is adopted for feature selection to reduce the data dimension. Next, the most valuable training samples are selected and labelled by human experts. The training samples are then fed into the classifier, while the Bayesian optimisation is applied to optimise the hyperparameters of the classification model. Finally, a set of experiments are performed to evaluate the performance of our method.

Keywords: industrial internet; intrusion detection; active learning; LightGBM.

DOI: 10.1504/IJCAT.2023.132095

International Journal of Computer Applications in Technology, 2023 Vol.71 No.3, pp.208 - 216

Received: 21 Feb 2022
Received in revised form: 04 Apr 2022
Accepted: 07 Apr 2022

Published online: 11 Jul 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article