concept bottleneck models
These models propose a structured way of learning where a model is forced to learn interpretable concepts as intermediate representations, advocating for improved interpretability and error analysis by aligning learned concepts with human-understandable categories.
- An Analysis of Concept Bottleneck Models: Measuring, Understanding, and Mitigating the Impact of Noisy Annotations
- Bayesian Concept Bottleneck Models with LLM Priors
- Deferring Concept Bottleneck Models: Learning to Defer Interventions to Inaccurate Experts
- DermaCon-IN: A Multiconcept-Annotated Dermatological Image Dataset of Indian Skin Disorders for Clinical AI Research
- Sample-efficient Learning of Concepts with Theoretical Guarantees: from Data to Concepts without Interventions
- Understanding and Improving Adversarial Robustness of Neural Probabilistic Circuits