/bioma-tcn-ae

Minimal Working Example of a (baseline) Temporal Convolutional Autoencoder (TCN-AE) for Anomaly Detection in Time Series

Primary LanguageJupyter Notebook

TCN-AE

Minimal working example of a (baseline) Temporal Convolutional Autoencoder (TCN-AE) for anomaly detection in time series, based on the paper:

Thill, Markus; Konen, Wolfgang; Bäck, Thomas (2020) Time Series Encodings with Temporal Convolutional Networks Inproceedings In: Vasile, Massimiliano; Filipic, Bogdan (Ed.): 9th International Conference on Bioinspired Optimisation Methods and Their Applications (BIOMA), Bruxelles.

This example was tested with the package versions specified in requirements.txt. We use the excellent keras-tcn package for our TCN blocks.

More details regarding the Mackey-Glass Anomaly Benchmark (MGAB) and it's source code can be found in the MGAB Repository.

More details to come...