The Annals of Applied Statistics
- Ann. Appl. Stat.
- Volume 4, Number 2 (2010), 645-662.
Bayesian anomaly detection methods for social networks
Learning the network structure of a large graph is computationally demanding, and dynamically monitoring the network over time for any changes in structure threatens to be more challenging still.
This paper presents a two-stage method for anomaly detection in dynamic graphs: the first stage uses simple, conjugate Bayesian models for discrete time counting processes to track the pairwise links of all nodes in the graph to assess normality of behavior; the second stage applies standard network inference tools on a greatly reduced subset of potentially anomalous nodes. The utility of the method is demonstrated on simulated and real data sets.
Ann. Appl. Stat., Volume 4, Number 2 (2010), 645-662.
First available in Project Euclid: 3 August 2010
Permanent link to this document
Digital Object Identifier
Mathematical Reviews number (MathSciNet)
Zentralblatt MATH identifier
Heard, Nicholas A.; Weston, David J.; Platanioti, Kiriaki; Hand, David J. Bayesian anomaly detection methods for social networks. Ann. Appl. Stat. 4 (2010), no. 2, 645--662. doi:10.1214/10-AOAS329. https://projecteuclid.org/euclid.aoas/1280842134
- Supplementary material A: Hurdle exponential family distributions. Details of the Bayesian inferential models considered in this paper.
- Supplementary material B: Matlab/Octave code. Matlab code written by DJW for implementing the models used in this paper.