Statistical Test for Attention Map in Vision Transformer

TL;DR

This paper introduces a statistical test for Vision Transformer attention maps, providing a rigorous way to assess the significance of attention regions, especially useful in high-stakes applications like medical diagnosis.

Contribution

It proposes a novel statistical testing framework for ViT attention maps using selective inference, enabling reliable significance assessment of attention regions.

Findings

Validates the method through numerical experiments

Demonstrates application in brain image diagnosis

Provides p-values for attention significance

Abstract

The Vision Transformer (ViT) demonstrates exceptional performance in various computer vision tasks. Attention is crucial for ViT to capture complex wide-ranging relationships among image patches, allowing the model to weigh the importance of image patches and aiding our understanding of the decision-making process. However, when utilizing the attention of ViT as evidence in high-stakes decision-making tasks such as medical diagnostics, a challenge arises due to the potential of attention mechanisms erroneously focusing on irrelevant regions. In this study, we propose a statistical test for ViT's attentions, enabling us to use the attentions as reliable quantitative evidence indicators for ViT's decision-making with a rigorously controlled error rate. Using the framework called selective inference, we quantify the statistical significance of attentions in the form of p-values, which enables the theoretically grounded quantification of the false positive detection probability of attentions. We demonstrate the validity and the effectiveness of the proposed method through numerical experiments and applications to brain image diagnoses.