Interactive Concept Bottleneck Models
This work addresses the need for more accurate and interpretable AI models in domains like medical imaging and bird classification by enabling interactive human-AI collaboration, though it is incremental as it builds on existing concept bottleneck models.
The paper tackles the problem of improving prediction accuracy in interactive settings by extending concept bottleneck models to query human collaborators for concept labels, achieving accuracy gains of 5-10% with only 5 interactions on datasets like Caltech-UCSD Birds, CheXpert, and OAI.
Concept bottleneck models (CBMs) are interpretable neural networks that first predict labels for human-interpretable concepts relevant to the prediction task, and then predict the final label based on the concept label predictions. We extend CBMs to interactive prediction settings where the model can query a human collaborator for the label to some concepts. We develop an interaction policy that, at prediction time, chooses which concepts to request a label for so as to maximally improve the final prediction. We demonstrate that a simple policy combining concept prediction uncertainty and influence of the concept on the final prediction achieves strong performance and outperforms static approaches as well as active feature acquisition methods proposed in the literature. We show that the interactive CBM can achieve accuracy gains of 5-10% with only 5 interactions over competitive baselines on the Caltech-UCSD Birds, CheXpert and OAI datasets.