Augmented Shortcuts for Vision Transformers

TL;DR

This paper addresses feature collapse in vision transformers by introducing augmented shortcuts with learnable parameters, improving feature diversity and accuracy without significant computational overhead.

Contribution

It proposes a novel augmented shortcut scheme with block-circulant projections to enhance feature diversity and accuracy in vision transformers.

Findings

Achieves about 1% accuracy improvement on benchmark datasets.

Effectively mitigates feature collapse in deep transformer models.

Maintains computational efficiency with the proposed method.

Abstract

Transformer models have achieved great progress on computer vision tasks recently. The rapid development of vision transformers is mainly contributed by their high representation ability for extracting informative features from input images. However, the mainstream transformer models are designed with deep architectures, and the feature diversity will be continuously reduced as the depth increases, i.e., feature collapse. In this paper, we theoretically analyze the feature collapse phenomenon and study the relationship between shortcuts and feature diversity in these transformer models. Then, we present an augmented shortcut scheme, which inserts additional paths with learnable parameters in parallel on the original shortcuts. To save the computational costs, we further explore an efficient approach that uses the block-circulant projection to implement augmented shortcuts. Extensive experiments conducted on benchmark datasets demonstrate the effectiveness of the proposed method, which brings about 1% accuracy increase of the state-of-the-art visual transformers without obviously increasing their parameters and FLOPs.