CB-SLICE: Concept-Based Interpretable Error Slice Discovery

Yael Konforti, Mateo Espinosa Zarlenga, Elaf Almahmoud, Mateja Jamnik
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:59949-59968, 2026.

Abstract

Despite strong average-case performance, deep learning models often exhibit systematic errors on specific population groups, known as error slices. Identifying these groups and the root causes of their failures is critical for model debugging and bias mitigation. However, existing error Slice Discovery Methods (SDMs) typically generate explanations disconnected from the model’s inference process, thus only approximating the underlying error source and may be inaccurate. We address this limitation by leveraging Concept Bottleneck Models (CBMs), whose predictions are directly dependent on human-understandable semantic concepts. Since downstream task failures in CBMs commonly arise from concept mispredictions, concept representations provide a strong candidate for error slice identification, offering fine-grained explanations directly linked to the error source. Building on this insight, we introduce CB-SLICE, a concept-based SDM that groups samples with shared concept prediction failures and identifies the keyword-concepts most responsible for each slice’s failure-mode. Across multiple benchmarks, we show that CB-SLICE outperforms state-of-the-art methods in uncovering well-known biases while providing richer and more faithful explanations of model errors.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-konforti26a, title = {{CB}-{SLICE}: Concept-Based Interpretable Error Slice Discovery}, author = {Konforti, Yael and Espinosa Zarlenga, Mateo and Almahmoud, Elaf and Jamnik, Mateja}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {59949--59968}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/konforti26a/konforti26a.pdf}, url = {https://proceedings.mlr.press/v306/konforti26a.html}, abstract = {Despite strong average-case performance, deep learning models often exhibit systematic errors on specific population groups, known as error slices. Identifying these groups and the root causes of their failures is critical for model debugging and bias mitigation. However, existing error Slice Discovery Methods (SDMs) typically generate explanations disconnected from the model’s inference process, thus only approximating the underlying error source and may be inaccurate. We address this limitation by leveraging Concept Bottleneck Models (CBMs), whose predictions are directly dependent on human-understandable semantic concepts. Since downstream task failures in CBMs commonly arise from concept mispredictions, concept representations provide a strong candidate for error slice identification, offering fine-grained explanations directly linked to the error source. Building on this insight, we introduce CB-SLICE, a concept-based SDM that groups samples with shared concept prediction failures and identifies the keyword-concepts most responsible for each slice’s failure-mode. Across multiple benchmarks, we show that CB-SLICE outperforms state-of-the-art methods in uncovering well-known biases while providing richer and more faithful explanations of model errors.} }
Endnote
%0 Conference Paper %T CB-SLICE: Concept-Based Interpretable Error Slice Discovery %A Yael Konforti %A Mateo Espinosa Zarlenga %A Elaf Almahmoud %A Mateja Jamnik %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-konforti26a %I PMLR %P 59949--59968 %U https://proceedings.mlr.press/v306/konforti26a.html %V 306 %X Despite strong average-case performance, deep learning models often exhibit systematic errors on specific population groups, known as error slices. Identifying these groups and the root causes of their failures is critical for model debugging and bias mitigation. However, existing error Slice Discovery Methods (SDMs) typically generate explanations disconnected from the model’s inference process, thus only approximating the underlying error source and may be inaccurate. We address this limitation by leveraging Concept Bottleneck Models (CBMs), whose predictions are directly dependent on human-understandable semantic concepts. Since downstream task failures in CBMs commonly arise from concept mispredictions, concept representations provide a strong candidate for error slice identification, offering fine-grained explanations directly linked to the error source. Building on this insight, we introduce CB-SLICE, a concept-based SDM that groups samples with shared concept prediction failures and identifies the keyword-concepts most responsible for each slice’s failure-mode. Across multiple benchmarks, we show that CB-SLICE outperforms state-of-the-art methods in uncovering well-known biases while providing richer and more faithful explanations of model errors.
APA
Konforti, Y., Espinosa Zarlenga, M., Almahmoud, E. & Jamnik, M.. (2026). CB-SLICE: Concept-Based Interpretable Error Slice Discovery. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:59949-59968 Available from https://proceedings.mlr.press/v306/konforti26a.html.

Related Material