Network Intrusion Detection Using CatBoost Algorithm

Vinay Jain

Network Intrusion Detection Using CatBoost Algorithm

Authors: Vinay Jain

Unique Paper ID: 148286
Volume: 6
Issue: 1
PageNo: 228-232

Keywords: Intrusion Detection NSL-KDD Dataset Boosting CatBoost algorithm Classification Accuracy False Positive Rate Detection Rate

Abstract:
With the rapid advancement in technology, the usage of devices generating digital data has surged and thus, resulted in increased network traffic. This has also raised the issues of network security commensurately as increased network traffic means increased vulnerability of data to hackers. Due to these reasons, intrusion detection system has been an important research issue. An intrusion detection system is like a defense mechanism that prevents unauthorized access to the data or network of an organization. Boosting algorithms are ensemble techniques which form a strong model from weak ones by taking into account the previous classifiers success. In this paper, an intrusion detection system is proposed using a boosting technique called CatBoost algorithm. A binary classification i.e., differentiating benign and malignant intrusions, and a multi-class classification i.e., identifying intrusions as benign or an attack type of the category DoS, Probe, U2R and R2L, is performed using CatBoost algorithm. Later the results from both types of classifications are analyzed to see the algorithmâ€™s efficiency in different detection scenarios.

Download article

email to a friend

Cite This Article

ISSN: 2349-6002
Volume: 6
Issue: 1
PageNo: 228-232

Network Intrusion Detection Using CatBoost Algorithm

Available:https://ijirt.org/Article?manuscript=148286

03 Jun, 2019

Comparative Analysis of Feature Selection Techniques for Network Intrusion Detection

Impact Factor
8.01 (Year 2024)

UGC Approved
Journal no 47859

Join Our IPN

IJIRT Partner Network

Submit your research paper and those of your network (friends, colleagues, or peers) through your IPN account, and receive 800 INR for each paper that gets published.

Join Now