# Alias-Free Convnets: Fractional Shift Invariance via Polynomial Activations

Source: [Alias-Free Convnets: Fractional Shift Invariance via Polynomial Activations](<https://www.sota2.com/research/paper/68fa6f921e209131cea46590>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-04T04:29:47.891Z

Authors: Hagay Michaeli, Tomer Michaeli, Daniel Soudry

Published: 2023-03-14T17:16:16.000Z

## 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.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Classification on ImageNet (val) (Adversarial Accuracy)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-val-adversarial-accuracy.md>): Image Classification; ImageNet (val); Baseline Accuracy: 81.04.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2303.08085>)

[PDF](<https://arxiv.org/pdf/2303.08085.pdf>)

[Implementation](<https://github.com/hmichaeli/alias_free_convnets/>)