Mutual Information on Low-rank Matrix for Effective Intrusion Detection

With the popularity of the internet expanding, security is a top concern nowadays. Computer networks are becoming more and more prone to cyberattacks and other dangers due to the massive volume of data flow in various formats. Finding significant information to detect intrusions becomes more difficu...

Full description

Saved in:
Bibliographic Details
Published in2022 4th International Conference on Sustainable Technologies for Industry 4.0 (STI) pp. 1 - 5
Main Authors Jannat, Hasin E, Rahman, Md Mahbubur, Dey, Samrat Kumar, Islam, A F M Mohaimenul
Format Conference Proceeding
LanguageEnglish
Published IEEE 17.12.2022
Subjects
Computer networks
Dimensionality reduction
Feature extraction
feature selection
Fourth Industrial Revolution
ids
Intrusion detection
low-rank matrix
Matrix decomposition
Security
Online AccessGet full text

Cover

More Information
Summary:With the popularity of the internet expanding, security is a top concern nowadays. Computer networks are becoming more and more prone to cyberattacks and other dangers due to the massive volume of data flow in various formats. Finding significant information to detect intrusions becomes more difficult as a result. In this data-driven world, dimensionality reduction is a crucial procedure for intrusion detection. For academics, an intriguing subject is choosing relevant traits from a big amount of irrelevant and associated data to accurately detect alerts/threats.One of the most popular statistical analysis-based methods for dimensionality reduction is principal component analysis (PCA), however, it has certain limitations. In this paper, we addressed this limitations of PCA and showed that mutual information of low-rank matrix is an effective subspace detection techniques as well as gives a better classification performance.
DOI:10.1109/STI56238.2022.10103298