Hierarchical, Interpretable, Label-Free Concept Bottleneck Model

TL;DR

HIL-CBM is a hierarchical, interpretable, label-free concept bottleneck model that improves interpretability and accuracy by mirroring human cognitive abstraction levels without needing relational annotations.

Contribution

It introduces a hierarchical framework for CBMs with a gradient-based loss and dual classification heads, enhancing interpretability and performance without relational concept annotations.

Findings

HIL-CBM outperforms state-of-the-art sparse CBMs in accuracy.

Human evaluations show HIL-CBM provides more interpretable explanations.

HIL-CBM maintains a hierarchical, label-free approach to feature concepts.

Abstract

Concept Bottleneck Models (CBMs) introduce interpretability to black-box deep learning models by predicting labels through human-understandable concepts. However, unlike humans, who identify objects at different levels of abstraction using both general and specific features, existing CBMs operate at a single semantic level in both concept and label space. We propose HIL-CBM, a Hierarchical Interpretable Label-Free Concept Bottleneck Model that extends CBMs into a hierarchical framework to enhance interpretability by more closely mirroring the human cognitive process. HIL-CBM enables classification and explanation across multiple semantic levels without requiring relational concept annotations. HIL-CBM aligns the abstraction level of concept-based explanations with that of model predictions, progressing from abstract to concrete. This is achieved by (i) introducing a gradient-based visual consistency loss that encourages abstraction layers to focus on similar spatial regions, and (ii) training dual classification heads, each operating on feature concepts at different abstraction levels. Experiments on benchmark datasets demonstrate that HIL-CBM outperforms state-of-the-art sparse CBMs in classification accuracy. Human evaluations further show that HIL-CBM provides more interpretable and accurate explanations, while maintaining a hierarchical and label-free approach to feature concepts.