Anomaly identification with limited sampling budget

Authors
Publication date 2016
Book title 2016 IEEE Information Theory Workshop (ITW 2016)
Book subtitle Cambridge, United Kingdom, 11-14 September 2016
ISBN
  • 9781509010912
ISBN (electronic)
  • 9781509010905
Series IEEE Conference Proceedings
Event Information Theory Workshop (ITW 2016)
Pages (from-to) 216-220
Publisher Piscataway, NJ: IEEE
Organisations
  • Faculty of Science (FNWI) - Korteweg-de Vries Institute for Mathematics (KdVI)
Abstract We consider a network of data streams from which an anomalous process with known target distribution is to be identified. Because in practice obtaining observations may be expensive, we assume that there is a constraint on the total number of observations based on which the decision has to be made. We derive a sufficient condition on the sampling budget such that the error probability is kept below some desired level. Furthermore, we show how to obtain a sampling allocation that can improve upon equal sampling allocation and achieves the desired accuracy.
Document type Conference contribution
Language English
Published at https://doi.org/10.1109/ITW.2016.7606827
Permalink to this page
Back