DCBM: Data-Efficient Visual Concept Bottleneck Models

Katharina Prasse, Patrick Knab, Sascha Marton, Christian Bartelt, Margret Keuper
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:49752-49782, 2025.

Abstract

Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts. However, current CBMs typically rely on concept sets extracted from large language models or extensive image corpora, limiting their effectiveness in data-sparse scenarios. We propose Data-efficient CBMs (DCBMs), which reduce the need for large sample sizes during concept generation while preserving interpretability. DCBMs define concepts as image regions detected by segmentation or detection foundation models, allowing each image to generate multiple concepts across different granularities. Exclusively containing dataset-specific concepts, DCBMs are well suited for fine-grained classification and out-of-distribution tasks. Attribution analysis using Grad-CAM demonstrates that DCBMs deliver visual concepts that can be localized in test images. By leveraging dataset-specific concepts instead of predefined or general ones, DCBMs enhance adaptability to new domains. The code is available at: https://github.com/KathPra/DCBM.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-prasse25a, title = {{DCBM}: Data-Efficient Visual Concept Bottleneck Models}, author = {Prasse, Katharina and Knab, Patrick and Marton, Sascha and Bartelt, Christian and Keuper, Margret}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {49752--49782}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/prasse25a/prasse25a.pdf}, url = {https://proceedings.mlr.press/v267/prasse25a.html}, abstract = {Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts. However, current CBMs typically rely on concept sets extracted from large language models or extensive image corpora, limiting their effectiveness in data-sparse scenarios. We propose Data-efficient CBMs (DCBMs), which reduce the need for large sample sizes during concept generation while preserving interpretability. DCBMs define concepts as image regions detected by segmentation or detection foundation models, allowing each image to generate multiple concepts across different granularities. Exclusively containing dataset-specific concepts, DCBMs are well suited for fine-grained classification and out-of-distribution tasks. Attribution analysis using Grad-CAM demonstrates that DCBMs deliver visual concepts that can be localized in test images. By leveraging dataset-specific concepts instead of predefined or general ones, DCBMs enhance adaptability to new domains. The code is available at: https://github.com/KathPra/DCBM.} }
Endnote
%0 Conference Paper %T DCBM: Data-Efficient Visual Concept Bottleneck Models %A Katharina Prasse %A Patrick Knab %A Sascha Marton %A Christian Bartelt %A Margret Keuper %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-prasse25a %I PMLR %P 49752--49782 %U https://proceedings.mlr.press/v267/prasse25a.html %V 267 %X Concept Bottleneck Models (CBMs) enhance the interpretability of neural networks by basing predictions on human-understandable concepts. However, current CBMs typically rely on concept sets extracted from large language models or extensive image corpora, limiting their effectiveness in data-sparse scenarios. We propose Data-efficient CBMs (DCBMs), which reduce the need for large sample sizes during concept generation while preserving interpretability. DCBMs define concepts as image regions detected by segmentation or detection foundation models, allowing each image to generate multiple concepts across different granularities. Exclusively containing dataset-specific concepts, DCBMs are well suited for fine-grained classification and out-of-distribution tasks. Attribution analysis using Grad-CAM demonstrates that DCBMs deliver visual concepts that can be localized in test images. By leveraging dataset-specific concepts instead of predefined or general ones, DCBMs enhance adaptability to new domains. The code is available at: https://github.com/KathPra/DCBM.
APA
Prasse, K., Knab, P., Marton, S., Bartelt, C. & Keuper, M.. (2025). DCBM: Data-Efficient Visual Concept Bottleneck Models. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:49752-49782 Available from https://proceedings.mlr.press/v267/prasse25a.html.

Related Material