Enhancing Concept Localization in CLIP-based Concept Bottleneck Models
This addresses explainable AI for users needing reliable concept-based explanations, but it is incremental as it builds on existing CLIP and CBM methods.
The paper tackles the problem of concept hallucination in CLIP-based Concept Bottleneck Models, which undermines explanation faithfulness, and introduces CHILI to localize concepts and generate more interpretable saliency explanations.
This paper addresses explainable AI (XAI) through the lens of Concept Bottleneck Models (CBMs) that do not require explicit concept annotations, relying instead on concepts extracted using CLIP in a zero-shot manner. We show that CLIP, which is central in these techniques, is prone to concept hallucination, incorrectly predicting the presence or absence of concepts within an image in scenarios used in numerous CBMs, hence undermining the faithfulness of explanations. To mitigate this issue, we introduce Concept Hallucination Inhibition via Localized Interpretability (CHILI), a technique that disentangles image embeddings and localizes pixels corresponding to target concepts. Furthermore, our approach supports the generation of saliency-based explanations that are more interpretable.