Learning Outlier Ensembles: The Best of Both Worlds–Supervised and Unsupervised

Research output: Contribution to conferencePaperResearchpeer-review

  • Barbora Micenková, ETH Zürich, Denmark
  • Brian McWilliams, ETH, Zurich, Switzerland
  • Ira Assent
Years of research in unsupervised outlier detection have produced numerous algorithms to score data according to their exceptionality. wever, the nature of outliers heavily depends on the application context and different algorithms are sensitive to outliers of different nature. This makes it very difficult to assess suitability of a particular algorithm without a priori knowledge. On the other hand, in many applications, some examples of outliers exist or can be obtain
edin addition to the vast amount of unlabeled data. Unfortunately, this extra knowledge cannot be simply incorporated into the existing unsupervised algorithms.
In this paper, we show how to use powerful machine learning approaches to combine labeled examples together with arbitrary unsupervised outlier scoring algorithms. We aim to get the best out of the two worlds—supervised and unsupervised. Our approach is also a viable solution to the recent problem of outlier ensemble selection.
Original languageEnglish
Publication year2014
Number of pages4
Publication statusPublished - 2014
EventACM SIGKDD 2014 Workshop ODD2 : Outlier Detection & Description under Data Diversity - New York, United States
Duration: 24 Aug 201424 Aug 2014
Conference number: 2


WorkshopACM SIGKDD 2014 Workshop ODD2
CountryUnited States
CityNew York

See relations at Aarhus University Citationformats

ID: 84415846