2019/12/04 by Gábor Horváth, Edith Kovács, Horváth, Gábor +5 · 2 citations
Computer Science · Physics and Astronomy · #Anomaly Detection Techniques and Applications #Complex Network Analysis Techniques #Computational Engineering #FOS: Computer and information sciences #FOS: Mathematics #Finance #I.2.1 #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Network Security and Intrusion Detection #Probability (math.PR) #and Science (cs.CE)
paper · pdf · doi:10.48550/arxiv.1912.02166
openalex publication_date 2019/12/04 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The anomaly detection method presented by this paper has a special feature:\nit does not only indicate whether an observation is anomalous or not but also\ntells what exactly makes an anomalous observation unusual. Hence, it provides\nsupport to localize the reason of the anomaly.\n The proposed approach is model-based; it relies on the multivariate\nprobability distribution associated with the observations. Since the rare\nevents are present in the tails of the probability distributions, we use copula\nfunctions, that are able to model the fat-tailed distributions well. The\npresented procedure scales well; it can cope with a large number of\nhigh-dimensional samples. Furthermore, our procedure can cope with missing\nvalues, too, which occur frequently in high-dimensional data sets.\n In the second part of the paper, we demonstrate the usability of the method\nthrough a case study, where we analyze a large data set consisting of the\nperformance counters of a real mobile telecommunication network. Since such\nnetworks are complex systems, the signs of sub-optimal operation can remain\nhidden for a potentially long time. With the proposed procedure, many such\nhidden issues can be isolated and indicated to the network operator.\n