Partially Interpretable Models with Guarantees on Coverage and Accuracy

Nave Frost, Zachary Lipton, Yishay Mansour, Michal Moshkovitz
Proceedings of The 35th International Conference on Algorithmic Learning Theory, PMLR 237:590-613, 2024.

Abstract

Simple, sufficient explanations furnished by short decision lists can be useful for guiding stakeholder actions. Unfortunately, this transparency can come at the expense of the higher accuracy enjoyed by black box methods, like deep nets. To date, practitioners typically either (i) insist on the simpler model, forsaking accuracy; or (ii) insist on maximizing accuracy, settling for post-hoc explanations of dubious faithfulness. In this paper, we propose a hybrid partially interpretable model that represents a compromise between the two extremes. In our setup, each input is first processed by a decision list that can either execute a decision or abstain, handing off authority to the opaque model. The key to optimizing the decision list is to optimally trade off the accuracy of the composite system against coverage (the fraction of the population that receives explanations). We contribute a new principled algorithm for constructing partially interpretable decision lists, providing theoretical guarantees addressing both interpretability and accuracy. As an instance of our result, we prove that when the optimal decision list has length $k$, coverage $c$, and $b$ mistakes, our algorithm will generate a decision list that has length no greater than $4k$, coverage at least $c/2$, and makes at most $4b$ mistakes. Finally, we empirically validate the effectiveness of the new model.

Cite this Paper

BibTeX
@InProceedings{pmlr-v237-frost24a, title = {Partially Interpretable Models with Guarantees on Coverage and Accuracy}, author = {Frost, Nave and Lipton, Zachary and Mansour, Yishay and Moshkovitz, Michal}, booktitle = {Proceedings of The 35th International Conference on Algorithmic Learning Theory}, pages = {590--613}, year = {2024}, editor = {Vernade, Claire and Hsu, Daniel}, volume = {237}, series = {Proceedings of Machine Learning Research}, month = {25--28 Feb}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v237/frost24a/frost24a.pdf}, url = {https://proceedings.mlr.press/v237/frost24a.html}, abstract = {Simple, sufficient explanations furnished by short decision lists can be useful for guiding stakeholder actions. Unfortunately, this transparency can come at the expense of the higher accuracy enjoyed by black box methods, like deep nets. To date, practitioners typically either (i) insist on the simpler model, forsaking accuracy; or (ii) insist on maximizing accuracy, settling for post-hoc explanations of dubious faithfulness. In this paper, we propose a hybrid partially interpretable model that represents a compromise between the two extremes. In our setup, each input is first processed by a decision list that can either execute a decision or abstain, handing off authority to the opaque model. The key to optimizing the decision list is to optimally trade off the accuracy of the composite system against coverage (the fraction of the population that receives explanations). We contribute a new principled algorithm for constructing partially interpretable decision lists, providing theoretical guarantees addressing both interpretability and accuracy. As an instance of our result, we prove that when the optimal decision list has length $k$, coverage $c$, and $b$ mistakes, our algorithm will generate a decision list that has length no greater than $4k$, coverage at least $c/2$, and makes at most $4b$ mistakes. Finally, we empirically validate the effectiveness of the new model.} }
Endnote
%0 Conference Paper %T Partially Interpretable Models with Guarantees on Coverage and Accuracy %A Nave Frost %A Zachary Lipton %A Yishay Mansour %A Michal Moshkovitz %B Proceedings of The 35th International Conference on Algorithmic Learning Theory %C Proceedings of Machine Learning Research %D 2024 %E Claire Vernade %E Daniel Hsu %F pmlr-v237-frost24a %I PMLR %P 590--613 %U https://proceedings.mlr.press/v237/frost24a.html %V 237 %X Simple, sufficient explanations furnished by short decision lists can be useful for guiding stakeholder actions. Unfortunately, this transparency can come at the expense of the higher accuracy enjoyed by black box methods, like deep nets. To date, practitioners typically either (i) insist on the simpler model, forsaking accuracy; or (ii) insist on maximizing accuracy, settling for post-hoc explanations of dubious faithfulness. In this paper, we propose a hybrid partially interpretable model that represents a compromise between the two extremes. In our setup, each input is first processed by a decision list that can either execute a decision or abstain, handing off authority to the opaque model. The key to optimizing the decision list is to optimally trade off the accuracy of the composite system against coverage (the fraction of the population that receives explanations). We contribute a new principled algorithm for constructing partially interpretable decision lists, providing theoretical guarantees addressing both interpretability and accuracy. As an instance of our result, we prove that when the optimal decision list has length $k$, coverage $c$, and $b$ mistakes, our algorithm will generate a decision list that has length no greater than $4k$, coverage at least $c/2$, and makes at most $4b$ mistakes. Finally, we empirically validate the effectiveness of the new model.
APA
Frost, N., Lipton, Z., Mansour, Y. & Moshkovitz, M.. (2024). Partially Interpretable Models with Guarantees on Coverage and Accuracy. Proceedings of The 35th International Conference on Algorithmic Learning Theory, in Proceedings of Machine Learning Research 237:590-613 Available from https://proceedings.mlr.press/v237/frost24a.html.

Related Material