Year: 2023
Type: article
Abstract: In order to protect computer networks from malicious activity, intrusion detection systems (IDS) are essential. Traditional rule-based IDS frequently experience significant false positive rates and struggle to keep up with changing attack techniques. This paper, utilizing the NSL-KDD dataset, suggests a machine learning strategy for intrusion detection to overcome these issues. The suggested method makes use of the capabilities of machine learning algorithms to efficiently identify and categorise network intrusions. The proposed model is trained on and tested against the NSL-KDD dataset, a benchmark dataset for intrusion detection research. The data was split into four feature subsets that were taken from the NSL-KDD dataset in order to evaluate the performance of these classifiers. Preprocessing techniques were used to remove pointless attributes from the dataset because an IDS's performance is influenced by the dimensions of the data. In this study, multiple machine learning (ML) classifiers were used to categorise data in an intrusion detection system (IDS) as either normal or invasive. The recommended machine learning-based IDS works brilliantly in terms of various accuracy parameters, according to testing results. When compared to traditional rule-based systems, the suggested method has improved detection rates and fewer false positives, improving the overall security of computer networks.
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7217358103
