AB-CF: Attention-based Counterfactual Explanation for Multivariate Time Series

This is the repository for our paper titled "Motif-guided time series counterfactual explanations". This paper has been accepted at the Big Data Analytics and Knowledge Discovery 25th International Conference, DaWaK 2023

Abstract

In this paper, we propose Attention-based Counterfactual Explanation (AB-CF), a novel model that generates post-hoc counterfactual explanations for multivariate time series classification that narrows the attention to a few important segments. We validated our model using seven real-world time-series datasets from the UEA repository. Our experimental results show the superiority of AB-CF in terms of validity, proximity, sparsity, contiguity, and efficiency compared with other competing state-of-the-art baselines.

Approach

Dataset

The data used in this project comes from the UEA archive.

Prerequisites

All python packages needed are listed in pip-requirements.txt file and can be installed simply using the pip command.

Reference

If you re-use this work, please cite:

@inproceedings{li2023attention, title={Attention-Based Counterfactual Explanation for Multivariate Time Series}, author={Li, Peiyu and Bahri, Omar and Boubrahimi, Souka{"\i}na Filali and Hamdi, Shah Muhammad}, booktitle={International Conference on Big Data Analytics and Knowledge Discovery}, pages={287--293}, year={2023}, organization={Springer} }