This was part of a capstone project for a major tech company. The dataset has not been included for security purposes, but the notebook shows how to use a recurrent neural network (RNN) to model time series data. With the modeled data, the fast generalized subset scan algorithm (FGSS, see http://www.jmlr.org/papers/volume14/mcfowland13a/mcfowland13a.pdf) is deployed to find and flag anomalous records and attributes.
danielw2904 / anomaly-detection-fgss- Goto Github PK
View Code? Open in Web Editor NEWThis project forked from ethanp55/anomaly-detection-fgss-
Anomaly detection using an RNN and the FGSS algorithm