One-Class Support Measure Machines for Group Anomaly Detection

Krikamol Muandet, Bernhard Schoelkopf
Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, PMLR R11:549-558, 2013.

Abstract

We propose one-class support measure ma- chines (OCSMMs) for group anomaly detec- tion. Unlike traditional anomaly detection, OCSMMs aim at recognizing anomalous ag- gregate behaviors of data points. The OC- SMMs generalize well-known one-class sup- port vector machines (OCSVMs) to a space of probability measures. By formulating the problem as quantile estimation on distribu- tions, we can establish interesting connec- tions to the OCSVMs and variable kernel density estimators (VKDEs) over the input space on which the distributions are defined, bridging the gap between large-margin meth- ods and kernel density estimators. In partic- ular, we show that various types of VKDEs can be considered as solutions to a class of regularization problems studied in this pa- per. Experiments on Sloan Digital Sky Sur- vey dataset and High Energy Particle Physics dataset demonstrate the benefits of the pro- posed framework in real-world applications.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR11-muandet13a, title = {One-Class Support Measure Machines for Group Anomaly Detection}, author = {Muandet, Krikamol and Schoelkopf, Bernhard}, booktitle = {Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence}, pages = {549--558}, year = {2013}, editor = {Nicholson, Ann and Smyth, Padhraic}, volume = {R11}, series = {Proceedings of Machine Learning Research}, month = {12--14 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r11/main/assets/muandet13a/muandet13a.pdf}, url = {https://proceedings.mlr.press/r11/muandet13a.html}, abstract = {We propose one-class support measure ma- chines (OCSMMs) for group anomaly detec- tion. Unlike traditional anomaly detection, OCSMMs aim at recognizing anomalous ag- gregate behaviors of data points. The OC- SMMs generalize well-known one-class sup- port vector machines (OCSVMs) to a space of probability measures. By formulating the problem as quantile estimation on distribu- tions, we can establish interesting connec- tions to the OCSVMs and variable kernel density estimators (VKDEs) over the input space on which the distributions are defined, bridging the gap between large-margin meth- ods and kernel density estimators. In partic- ular, we show that various types of VKDEs can be considered as solutions to a class of regularization problems studied in this pa- per. Experiments on Sloan Digital Sky Sur- vey dataset and High Energy Particle Physics dataset demonstrate the benefits of the pro- posed framework in real-world applications.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T One-Class Support Measure Machines for Group Anomaly Detection %A Krikamol Muandet %A Bernhard Schoelkopf %B Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2013 %E Ann Nicholson %E Padhraic Smyth %F pmlr-vR11-muandet13a %I PMLR %P 549--558 %U https://proceedings.mlr.press/r11/muandet13a.html %V R11 %X We propose one-class support measure ma- chines (OCSMMs) for group anomaly detec- tion. Unlike traditional anomaly detection, OCSMMs aim at recognizing anomalous ag- gregate behaviors of data points. The OC- SMMs generalize well-known one-class sup- port vector machines (OCSVMs) to a space of probability measures. By formulating the problem as quantile estimation on distribu- tions, we can establish interesting connec- tions to the OCSVMs and variable kernel density estimators (VKDEs) over the input space on which the distributions are defined, bridging the gap between large-margin meth- ods and kernel density estimators. In partic- ular, we show that various types of VKDEs can be considered as solutions to a class of regularization problems studied in this pa- per. Experiments on Sloan Digital Sky Sur- vey dataset and High Energy Particle Physics dataset demonstrate the benefits of the pro- posed framework in real-world applications. %Z Reissued by PMLR on 04 October 2026.
APA
Muandet, K. & Schoelkopf, B.. (2013). One-Class Support Measure Machines for Group Anomaly Detection. Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R11:549-558 Available from https://proceedings.mlr.press/r11/muandet13a.html. Reissued by PMLR on 04 October 2026.

Related Material