A new intelligent intrusion detector based on ensemble of decision trees

Artificial intelligence and machine learning are in widespread use nowadays in order to develop automatic and precise models for different tasks especially in the Internet. In this paper, by the use of machine learning techniques, an intrusion detection system is proposed. An intrusion detection sys...

Full description

Saved in:
Bibliographic Details
Published in:Journal of ambient intelligence and humanized computing 2022-07, Vol.13 (7), p.3347-3359
Main Authors: Mousavi, Seyed Morteza, Majidnezhad, Vahid, Naghipour, Avaz
Format: Article
Language:English
Subjects:
Accuracy
Algorithms
Ant colony optimization
Artificial Intelligence
Computational Intelligence
Correlation coefficients
Decision trees
Efficiency
Engineering
Intrusion detection systems
Machine learning
Neural networks
Original Research
Robotics and Automation
Support vector machines
User Interfaces and Human Computer Interaction
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Artificial intelligence and machine learning are in widespread use nowadays in order to develop automatic and precise models for different tasks especially in the Internet. In this paper, by the use of machine learning techniques, an intrusion detection system is proposed. An intrusion detection system is involved extensive mass of data; such data is naturally characterized with repetitions and noise which leads to the reduction in the stability and the accuracy of the intrusion detection system. Hence, the issue of reducing features dimensions for achieving a smaller subset of features which can precisely express the results and status of network observations has attracted a lot of researchers’ attention. In the proposed method, by using gradually feature removal method, 16 critical features were selected for representing various network visits. By combining ant colony algorithm and ensemble of decision trees, we proposed an efficient and stable classifier for judging a network visit to be normal or not. Despite the selection of 16 features, high accuracy, i.e. 99.92%, and the average value of Matthews correlation coefficient 0.91 are obtained.
ISSN:1868-5137
1868-5145
DOI:10.1007/s12652-019-01596-5