Volume 9, Number 9, July 2019

Ensemble learning using frequent itemset mining for anomaly detection

  Authors

Saeid Soheily-Khah and Yiming Wu, SKYLADS Research Team, France

  Abstract

Anomaly detection is vital for automated data analysis, with specific applications spanning almost every domain. In this paper, we propose a hybrid supervised learning of anomaly detection using frequent itemset mining and random forest with an ensemble probabilistic voting method, which outperforms the alternative supervised learning methods through the commonly used measures for anomaly detection: accuracy, true positive rate (i.e. recall) and false positive rate. To justify our claim, a benchmark dataset is used to evaluate the efficiency of the proposed approach, where the results illustrate its benefits.

  Keywords

Ensemble learning, anomaly detection, frequent (closed / maximal) itemset mining, random forest, classification

Conference Proceedings