My work currently revolves around using explainable AI methods to improve our ability to do science. Most recently, I have developped Quantitative Attributions with Counterfactuals (Adjavon et al., 2024) or QuAC for short. This is a method that creates counterfactual explanations to describe the difference between classes, quantified by how well the features found affect the decisions of a pre-trained classifier.
References
2024
Quantitative Attributions with Counterfactuals
Diane-Yayra Adjavon, Nils Eckstein, Alexander S. Bates, and 2 more authors
@article{adjavon2024quantitative,title={Quantitative Attributions with Counterfactuals},author={Adjavon, Diane-Yayra and Eckstein, Nils and Bates, Alexander S. and Jefferis, Gregory S.X.E. and Funke, Jan},year={2024},month=dec,doi={10.1101/2024.11.26.625505},url={https://www.biorxiv.org/content/10.1101/2024.11.26.625505v1.full.pdf},}