Naslov
Hybrid Data Mining Approaches for Intrusion Detection in the Internet of Things

Autori
Oreški, Dijana ; Andročec, Darko

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of International Conference on Smart Systems and Technologies 2018 (SST 2018) / Žagar, Drago ; Martinović, Goran ; Rimac Drlje, Snježana ; Galić, Irena - Osijek : Faculty of Electrical Engineering, Computer Science and Information Technology, 2018, 221-226

ISBN
978-1-5386-7189-4

Skup
International Conference on Smart Systems and Technologies 2018 (SST 2018)

Mjesto i datum
Osijek, Hrvatska, 10.10.2018. - 12.10.2018

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
intrusion detection ; data mining ; Internet of things ; feature selection ; security

Sažetak
Internet of things devices and services are often not designed with security in mind. For this reason, malicious users can create botnets and other malicious software targeting things’ vulnerabilities. In this work, we have tested various data mining techniques and proposed one that gives representing intrusion detection results with small percentage of false positives. Development of a successful prediction model largely depends on data preprocessing phase. Feature reduction implemented as feature extraction or feature selection is main step of preprocessing phase. This paper compares the applications of principal component analysis as feature extraction method and Relief, Information Gain, Gini Index and SfFS as feature selection methods to reduce features for decision tree classification. By examining NSL-KDD data set, the experiment shows that decision trees by feature selection using SfFS can perform significantly better than other approaches.

Izvorni jezik
Engleski

Znanstvena područja
Informacijske i komunikacijske znanosti

POVEZANOST RADA