Comparison of Ensemble Learning Methods Applied to Network Intrusion Detection
| Title | Comparison of Ensemble Learning Methods Applied to Network Intrusion Detection |
| Publication Type | Conference Paper |
| Year of Publication | 2017 |
| Authors | Belouch, Mustapha, hadaj, Salah El |
| Conference Name | Proceedings of the Second International Conference on Internet of Things, Data and Cloud Computing |
| Publisher | ACM |
| Conference Location | New York, NY, USA |
| ISBN Number | 978-1-4503-4774-7 |
| Keywords | Bagging, boosting, composability, ensemble classifiers, False Data Detection, Intrusion detection, Metrics, network intrusion detection system, pubcrawl, resilience, Resiliency, Stacking |
| Abstract | This paper investigates the possibility of using ensemble learning methods to improve the performance of intrusion detection systems. We compare an ensemble of three ensemble learning methods, boosting, bagging and stacking in order to improve the detection rate and to reduce the false alarm rate. These ensemble methods use well-known and different base classification algorithms, J48 (decision tree), NB (Naive Bayes), MLP (Neural Network) and REPTree. The comparison experiments are applied on UNSW-NB15 data set a recent public data set for network intrusion detection systems. Results show that using boosting, bagging can achieve higher accuracy than single classifier but stacking performs better than other ensemble learning methods. |
| URL | http://doi.acm.org/10.1145/3018896.3065830 |
| DOI | 10.1145/3018896.3065830 |
| Citation Key | belouch_comparison_2017 |