Code release for Interpreting Deep Models through the Lens of Data at ICJNN 2020
Paper:
https://arxiv.org/abs/2005.03442
https://doi.org/10.1109/IJCNN48605.2020.9207704
The used datasets are publically available:
Anomaly detection dataset: https://bit.ly/2UNk0Lo
FordB and Character Trajectories: http://www.timeseriesclassification.com/index.php
Use the main.py file to execute a selected procedure or the execute.sh file for a complete analysis. The .ipynb files provide additional methods to perform a post-hoc analysis using the trained models.