Alias-Free Convnets: Fractional Shift Invariance via Polynomial Activations

Hagay Michaeli, Tomer Michaeli, Daniel Soudry
Technion
CVPR 2023
Paper Code arXiv CVPR

Abstract

Although CNNs are believed to be invariant to translations, recent works have shown this is not the case, due to aliasing effects that stem from downsampling layers. The existing architectural solutions to prevent aliasing are partial since they do not solve these effects, that originate in non-linearities. We propose an extended anti-aliasing method that tackles both downsampling and non-linear layers, thus creating truly alias-free, shift-invariant CNNs. We show that the presented model is invariant to integer as well as fractional (i.e., sub-pixel) translations, thus outperforming other shift-invariant methods in terms of robustness to adversarial translations.

Proposed Model

My Happy SVG

Results

Image 1 Image 2 Image 1 Image 2
Perfect shift-invariance
(circular shifts, integer and fractional) 		Improved robustness
("Crop-shifts")

CVPR Presentation

Slides

Poster

BibTeX


      @InProceedings{Michaeli_2023_CVPR,
    author    = {Michaeli, Hagay and Michaeli, Tomer and Soudry, Daniel},
    title     = {Alias-Free Convnets: Fractional Shift Invariance via Polynomial Activations},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2023},
    pages     = {16333-16342}
}