Alias-Free ViT: Fractional Shift Invariance via Linear Attention

TL;DR

This paper introduces an Alias-Free Vision Transformer that employs alias-free downsampling and linear attention to achieve shift invariance, enhancing robustness to image translations while maintaining competitive classification performance.

Contribution

It proposes a novel ViT architecture combining alias-free components and linear attention for improved translation robustness.

Findings

Outperforms similar models in adversarial translation robustness

Maintains competitive accuracy in image classification

Achieves shift invariance through novel architectural components

Abstract

Transformers have emerged as a competitive alternative to convnets in vision tasks, yet they lack the architectural inductive bias of convnets, which may hinder their potential performance. Specifically, Vision Transformers (ViTs) are not translation-invariant and are more sensitive to minor image translations than standard convnets. Previous studies have shown, however, that convnets are also not perfectly shift-invariant, due to aliasing in downsampling and nonlinear layers. Consequently, anti-aliasing approaches have been proposed to certify convnets' translation robustness. Building on this line of work, we propose an Alias-Free ViT, which combines two main components. First, it uses alias-free downsampling and nonlinearities. Second, it uses linear cross-covariance attention that is shift-equivariant to both integer and fractional translations, enabling a shift-invariant global representation. Our model maintains competitive performance in image classification and outperforms similar-sized models in terms of robustness to adversarial translations.