CausalConceptTS: Causal Attributions for Time Series Classification using High Fidelity Diffusion Models

This is the official repository for the paper CausalConceptTS: Causal Attributions for Time Series Classification using High Fidelity Diffusion Models

In this study, within the context of time series classification, we introduce a novel framework to assess the causal effect of concepts, i.e., predefined segments within a time series, on specific classification outcomes. To achieve this, we leverage state-of-the-art diffusion-based generative models to estimate counterfactual outcomes.

Results

We prove our approach efficace through three tasks:

• Drought prediction

• ECG classification

• EEG classification

Experiments

• Download the data from this link

• Place the desired test set under the data directory

• Follow the instructions under demo.ipynb to obtain the causal effects.

We welcome contributions to improve the reproducibility of this project! Feel free to submit pull requests or open issues.

Reference

@misc{alcaraz2024causalconceptts,
      title={CausalConceptTS: Causal Attributions for Time Series Classification using High Fidelity Diffusion Models}, 
      author={Juan Miguel Lopez Alcaraz and Nils Strodthoff},
      year={2024},
      eprint={2405.15871},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}