Anomaly Detection in Networks with Application to Financial Transaction Networks

This page contains additional content for the paper "Anomaly Detection in Networks with Application to Financial Transaction Networks" which is a joint work between The Alan Turing Institute and Accenture Plc, funded by the Accenture-Turing Alliance.

Paper + Supplementary Information

The paper can found here.

Software

As part of this work we will release two separate set of code. The first is an easy to use version which was constructed to demonstrate the methods in this paper and can be found here and is released under a GPLv3 license. The method has been modified to allow them to run relatively quickly on a small machine, removing a small number of statistics, and modifying constants to decrease computational requirements.

The second piece of software we release of the exact code used to generate the results in the paper which can be found here which we release for reproducibility. We note that this code is harder to use, as it uses an older version of Python, and an older version of certain libraries. To quickly test a new network we advise using the first set of code, for reproducibility this version.

If you have any problems or find any bugs please email the corresponding author on the paper.