On the Suitability of SHAP Explanations for Refining Classifications

Yusuf Arslan, Bertrand Lebichot, Kevin Allix, Lisa Veiber, Clément Lefebvre, Andrey Boytsov, Anne Goujon, Tegawendé Bissyande, Jacques Klein

2022

Abstract

In industrial contexts, when an ML model classifies a sample as positive, it raises an alarm, which is subsequently sent to human analysts for verification. Reducing the number of false alarms upstream in an ML pipeline is paramount to reduce the workload of experts while increasing customers’ trust. Increasingly, SHAP Explanations are leveraged to facilitate manual analysis. Because they have been shown to be useful to human analysts in the detection of false positives, we postulate that SHAP Explanations may provide a means to automate false-positive reduction. To confirm our intuition, we evaluate clustering and rules detection metrics with ground truth labels to understand the utility of SHAP Explanations to discriminate false positives from true positives. We show that SHAP Explanations are indeed relevant in discriminating samples and are a relevant candidate to automate ML tasks and help to detect and reduce false-positive results.

Paper Citation

in Harvard Style

Arslan Y., Lebichot B., Allix K., Veiber L., Lefebvre C., Boytsov A., Goujon A., Bissyande T. and Klein J. (2022). On the Suitability of SHAP Explanations for Refining Classifications. In Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART, ISBN 978-989-758-547-0, pages 395-402. DOI: 10.5220/0010827700003116

in Bibtex Style

@conference{icaart22,
author={Yusuf Arslan and Bertrand Lebichot and Kevin Allix and Lisa Veiber and Clément Lefebvre and Andrey Boytsov and Anne Goujon and Tegawendé Bissyande and Jacques Klein},
title={On the Suitability of SHAP Explanations for Refining Classifications},
booktitle={Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART,},
year={2022},
pages={395-402},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010827700003116},
isbn={978-989-758-547-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART,
TI - On the Suitability of SHAP Explanations for Refining Classifications
SN - 978-989-758-547-0
AU - Arslan Y.
AU - Lebichot B.
AU - Allix K.
AU - Veiber L.
AU - Lefebvre C.
AU - Boytsov A.
AU - Goujon A.
AU - Bissyande T.
AU - Klein J.
PY - 2022
SP - 395
EP - 402
DO - 10.5220/0010827700003116