English
 
Help Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT
Add to Basket
  DetailsSummary

Released
Conference Paper

Maximally divergent intervals for anomaly detection

MPS-Authors
/persons/resource/persons195348

Flach,  Milan
Department Biogeochemical Integration, Dr. M. Reichstein, Max Planck Institute for Biogeochemistry, Max Planck Society;

/persons/resource/persons62472

Mahecha,  Miguel D.
Department Biogeochemical Integration, Dr. M. Reichstein, Max Planck Institute for Biogeochemistry, Max Planck Society;
Michael Stifel Center for Data-driven and Simulation Science, Jena, Germany;
Empirical Inference of the Earth System, Dr. Miguel D. Mahecha, Department Biogeochemical Integration, Dr. M. Reichstein, Max Planck Institute for Biogeochemistry, Max Planck Society;

/persons/resource/persons197682

Bodesheim,  Paul
Department Biogeochemical Integration, Dr. M. Reichstein, Max Planck Institute for Biogeochemistry, Max Planck Society;

/persons/resource/persons62524

Reichstein,  Markus
Department Biogeochemical Integration, Dr. M. Reichstein, Max Planck Institute for Biogeochemistry, Max Planck Society;
Michael Stifel Center for Data-driven and Simulation Science, Jena, Germany;

External Resource

http://dx.doi.org/10.17871/BACI_ICML2016_Rodner
(Any fulltext)

Fulltext (restricted access)
There are currently no full texts shared for your IP range.
Fulltext (public)

BGC2496.pdf
(Publisher version), 384KB

Supplementary Material (public)
There is no public supplementary material available
Citation

Rodner, E., Barz, B., Guanche, Y., Flach, M., Mahecha, M. D., Bodesheim, P., et al. (2016). Maximally divergent intervals for anomaly detection. In ICML 2016 Anomaly Detection Workshop. doi:10.17871/BACI_ICML2016_Rodner.


Cite as: https://hdl.handle.net/11858/00-001M-0000-002B-4F5E-B
Abstract
We present new methods for batch anomaly detection in multivariate time series. Our methods are based on maximizing the Kullback-Leibler divergence between the data distribution within and outside an interval of the time series. An empirical analysis shows the benefits of our algorithms compared to methods that treat each time step independently from each other without optimizing with respect to all possible intervals.