Tell me why: Visual foundation models as self-explainable classifiers

TL;DR

This paper introduces ProtoFM, a lightweight, self-explainable classification method combining visual foundation models with a prototypical architecture, achieving competitive accuracy and improved interpretability.

Contribution

It presents a novel approach that integrates VFMs with a prototypical architecture and specialized training to produce efficient, interpretable classifiers with faithful explanations.

Findings

Achieves competitive classification performance.

Outperforms existing models on interpretability metrics.

Uses only a lightweight head on frozen VFMs.

Abstract

Visual foundation models (VFMs) have become increasingly popular due to their state-of-the-art performance. However, interpretability remains crucial for critical applications. In this sense, self-explainable models (SEM) aim to provide interpretable classifiers that decompose predictions into a weighted sum of interpretable concepts. Despite their promise, recent studies have shown that these explanations often lack faithfulness. In this work, we combine VFMs with a novel prototypical architecture and specialized training objectives. By training only a lightweight head (approximately 1M parameters) on top of frozen VFMs, our approach (ProtoFM) offers an efficient and interpretable solution. Evaluations demonstrate that our approach achieves competitive classification performance while outperforming existing models across a range of interpretability metrics derived from the literature. Code is available at https://github.com/hturbe/proto-fm.