Concept Bottleneck Models

Also known as CBM

Overview

Concept Bottleneck Models predict through an intermediate set of meaningful attributes, such as visual features or clinical findings, so people can examine how those concepts contribute to an output. Approaches include supervised concept predictors with simple classifiers, vision-language models that propose concepts, and structured bottlenecks that model concept relations, hierarchies, or spatial locations. The methods are tested in settings such as fine-grained image classification, medical diagnosis, text and graph prediction, and multimodal tasks, with evaluation increasingly examining concept accuracy and interventions as well as task accuracy. A central tension is that incomplete or noisy concepts can hurt predictions, while information leakage can preserve accuracy but make explanations or user interventions unreliable.

Latest papers

CardsList

92 papers9 papers in the last four weeks, down 18% on the four weeks before. 0.1% of all new papers.

  1. When Covariance Is Not Enough: Kernel-Based Regularization for Single-Cell Counterfactuals

    Nov 10, 2026Korbinian Traeuble, Edward de Brouwer, Archit Verma +2hilbert-schmidt independence criterionCausal Counterfactual Generation

  2. A Concept-Based Interpretable Framework for Clinical Text-based Prediction via Token-Concept Attention without Concept Supervision

    Nov 10, 2026Jifan Gao, Junjie Hu, John Caskey +6Concept-Based ExplanationsConcept Bottleneck Models

  3. Predictive Surprise as Self-Grounding Concept Bottleneck for Interpretable Time Series

    Nov 7, 2026Sachith Abeywickrama, Emadeldeen Eldele, Min Wu +2Interpretable Time-Series ClassificationConcept Bottleneck Models

  4. PriCBMs: Multi-granular Concept Bottleneck Vision-Language Models for Visual Privacy Recognition

    Nov 2, 2026Shenglei Pei, Haotian Sun, Pengfei Li +2Concept Bottleneck ModelsImage Classification

  5. Interpretable and Steerable Molecular Graph Generation using Concept Bottlenecks

    Oct 25, 2026Sarabeshwar Balaji, Ritesh Kumar, Lav R. VarshneyMolecular Graph GenerationMolecular Generation

  6. Learning to Measure and Diagnose: A Clinical-Dictionary Radiomic Bottleneck Model for Thyroid Nodule Classification

    Oct 21, 2026Guibin Zhang, Weier Luo, Lei CaoThyroid Nodule DetectionMedical Image Classification

  7. Auditing Where Concept Bottleneck Accuracy Comes From on CUB

    Oct 20, 2026Sascha Saralajew, Eleonora Poeta, Eliana Pastor +1Concept Bottleneck ModelsVisual Grounding

  8. Residual Stochastic Concept Bottleneck Models

    Oct 16, 2026Stephen Marcou, Mateo Espinosa Zarlenga, Sonia Laguna +3Concept Bottleneck Models

  9. Harmful Partial Interventions and Benign Leakage in Concept-based Models

    Oct 15, 2026Gregory Verghese, Christopher J. Soelistyo, Mateo Espinosa Zarlenga +2Concept Bottleneck Models

  10. Beyond Explanation: Debugging Medical Imaging Models via Concept Intervention

    Oct 6, 2026Samrajya Thapa, Daniel J. Quest, Timothy L. Kline +3Explainable Medical Image AnalysisConcept Bottleneck Models

  11. AnyBottle: A Recipe to Only Keep the Concepts You Really Need

    Oct 6, 2026Wolfgang Stammer, Sukrut Rao, Hevra Petekkaya +2Efficient Neural Network InferenceConcept Bottleneck Models

  12. When Are Concept Bottleneck Model Explanations Faithful and Compact?

    Oct 5, 2026Stefano Teso, Emanuele Marconato, Steve Azzolin +1Explainable Artificial IntelligenceConcept Bottleneck Models

  13. From Speech to Editable Concepts: Probing Emotion Recognition with Concept Bottleneck Models

    Sep 30, 2026Hezhao Zhang, Thomas HainConcept Bottleneck ModelsEmotion Recognition in Conversations

  14. Interpretable but Fragile? Robustness of Concept Bottlenecks under Geometric-Semantic Perturbations

    Sep 29, 2026Hanwei Zhang, Tianma Hu, Gaojie Jin +2Concept Bottleneck ModelsCertified Robustness

  15. Aperture: Training-Free Multiscale Concept Bottlenecks for Remote Sensing

    Sep 29, 2026Rishabh Mondal, Nipun Batra, Utkarsh MallRemote Sensing Image UnderstandingConcept Bottleneck Models

  16. CHOQOLATE: Organizing Concept Bottleneck Latent Spaces with Choquet Integrals

    Sep 29, 2026Rémi Kazmierczak, Johanne Cohen, Marianne ClauselConcept Bottleneck ModelsVLM Interpretability

  17. DF-CBM: Region-Aware Concept Bottleneck Models for Deepfake Detection

    Sep 28, 2026Georgios Tsoumplekas, Vazgken Vanian, Alexandros Doumanoglou +4Concept Bottleneck ModelsExplainable Deepfake Detection

  18. TRACE: Tractable Routing Autoencoder for Clinical ECG

    Sep 20, 2026Shunbo Jia, Runze Ma, Haonan Lyu +3Concept Bottleneck ModelsAutoencoders

  19. When Do Language-Grounded Explanations Help? A Graph-Bottleneck for Farm Monitoring Interpretable Sheep Facial Pain

    Sep 17, 2026Alam Noor, Miguel Guti'errez Gait'anConcept Bottleneck ModelsFacial Expression Recognition

  20. How to make effective use of domain experts for image classification?

    Sep 15, 2026Dieu-Donné Fangnon, Diane Lingrand, Aurélie Liard +3Image ClassificationConcept Bottleneck Models

  21. Concept-Grounded Reasoning with Prompt-Driven Localization for Interpretable Structured Report Generation

    Sep 14, 2026Xinyue Xu, Hongbin Lin, Juangui Xu +5Radiology Report GenerationConcept Bottleneck Models

  22. Explainable Prediction from Mobile Sensing Data through LLM-guided Concept Integration

    Sep 9, 2026Yuning Wang, Iman Azimi, Amir M. Rahmani +1Concept Bottleneck ModelsInterpretable ML

  23. Are Concept Bottleneck Models Effective as Decision-Support Systems?

    Aug 26, 2026Alessandro Bogani, Nicola Debole, Emanuele Marconato +3Concept Bottleneck ModelsHuman-AI Decision Making

  24. Weakly supervised concept Bottleneck Learning for Robust Two stage Object centric visual reasoning

    Aug 23, 2026Sparsh Tiwari, Gesina Schwalbe, Bettina FinzelConcept Bottleneck ModelsObject-Centric Representation Learning

  25. ConceptTS: LLM-Guided Concept Bottlenecks for Interpretable Multivariate Time-Series Forecasting

    Aug 21, 2026Yichen Jiang, Yueqiao Chen, Dongyu LiuMultivariate Time Series ForecastingExplainable Time Series Forecasting

  26. TRACE: Training-time Report-guided and Clinically Ordered Concept Editing

    Aug 21, 2026Wentao Yue, Tianyou Lai, Jiayu Luo +4Concept Bottleneck ModelsMedical Image Classification