Hierarchical Probabilistic Models for Group Anomaly Detection (2011)

Liang Xiong, Barnabas Poczos, Jeff Schneider, Andrew Connolly, Jake VanderPlas

Tags

group anomaly detection, topic models

Abstract

  Statistical anomaly detection typically focuses on finding
  individual data point anomalies. Often the most interesting or
  unusual things in a data set are not odd individual points, but
  rather larger scale phenomena that only become apparent when groups
  of data points are considered. In this paper, we propose two
  hierarchical probabilistic models for detecting such group
  anomalies. We evaluate our methods on synthetic data as well as
  astronomical data from the Sloan Digital Sky Survey. The
  experimental results show that the proposed models are effective in
  detecting group anomalies.

Full text

Download (application/pdf, 1.8 MB)

Approximate BibTeX Entry

@inproceedings{lx:11:mgm,
    Month = {April},
    Year = {2011},
    Booktitle = {AISTATS 2011},
    Author = { Liang Xiong, Barnabas Poczos, Jeff Schneider, Andrew Connolly, Jake VanderPlas },
    Title = {Hierarchical Probabilistic Models for Group Anomaly Detection}
}