Network intrusion detection through online transformation of eigenvector reflecting concept drift

Seongchul Park, Sanghyun Seo, Changhoon Jeong, Juntae Kim

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations

Abstract

Recently, large amount data streams are increasing. It is difficult to continuously store data and perform the principal component analysis in periodical offline (batch) mode. To solve this problem, there is a need to reflect the concept drift through online transformation to obtain the eigenvector, which is the goal of the principal component analysis. In this study, we compared the performance of online mode using the online eigenvector transformation in the network intrusion detection with offline mode. Both of them are applied through a multinomial logistic regression(MLR). The results showed that both the online and offline mode demonstrated excellent performance in accuracy, but the multinomial logistic regression applied with the online eigenvector transformation showed better performance in recall and as a result, the F1-Measure was also better.

Original languageEnglish
Title of host publicationProceedings of the 1st International Conference on Data Science, E-Learning and Information Systems, DATA 2018
PublisherAssociation for Computing Machinery
ISBN (Electronic)9781450365369
DOIs
StatePublished - 1 Oct 2018
Event1st International Conference on Data Science, E-Learning and Information Systems, DATA 2018 - Madrid, Spain
Duration: 1 Oct 20182 Oct 2018

Publication series

NameACM International Conference Proceeding Series

Conference

Conference1st International Conference on Data Science, E-Learning and Information Systems, DATA 2018
Country/TerritorySpain
CityMadrid
Period1/10/182/10/18

Keywords

  • Eigenvalue
  • Eigenvector
  • Online
  • PCA
  • Principle Component Analysis
  • Transformation

Fingerprint

Dive into the research topics of 'Network intrusion detection through online transformation of eigenvector reflecting concept drift'. Together they form a unique fingerprint.

Cite this