Interactive Concept Bottleneck Models

TL;DR

This paper introduces an interactive extension to concept bottleneck models that actively queries human labels for concepts to improve prediction accuracy, demonstrating significant gains with minimal interactions.

Contribution

It develops an interaction policy for CBMs that adaptively queries concepts, outperforming static and existing active feature acquisition methods.

Findings

Achieves 5-10% accuracy improvements with 5 interactions

Outperforms static approaches and existing active acquisition methods

Effective across multiple datasets including Caltech-UCSD Birds, CheXpert, and OAI

Abstract

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.