mlsanityData to 14 Oct 2026

Alias-Free Convnets: Fractional Shift Invariance via Polynomial Activations

arXiv 2303.08085 ↗First posted 14 Mar 2023cs.CV eess.IVcited by 11

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Builds on 22 of its 36 references that are in the map. First posted on arXiv 14 Mar 2023.

Builds on

22 of 36 references.

References · 22

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  3. 2022A ConvNet for the 2020scited by 3,579
  4. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,877
  5. 2021Alias-Free Generative Adversarial Networkscited by 1,184
  6. 2016Geometric deep learning: going beyond Euclidean datacited by 1,048
  7. 2019General E(2)-Equivariant Steerable CNNscited by 446
  8. 2017Joint 2D-3D-Semantic Data for Indoor Scene Understandingcited by 406
  9. 2019Making Convolutional Networks Shift-Invariant Againcited by 391
  10. 2019Universal Approximation with Deep Narrow Networkscited by 200
  11. 2018Why do deep convolutional networks generalize so poorly to small image transformations?cited by 198
  12. 2020Attentive Group Equivariant Convolutional Networkscited by 63
10 more
  1. 2020Delving Deeper into Anti-aliasing in ConvNetscited by 60
  2. 2020Truly shift-invariant convolutional neural networkscited by 53
  3. 2020An Effective Anti-Aliasing Approach for Residual Networkscited by 23
  4. 2022Equivariance Discovery by Learned Parameter-Sharingcited by 18
  5. 2022Learnable Polyphase Sampling for Shift Invariant and Equivariant Convolutional Networkscited by 18
  6. 2021Shift Invariance Can Reduce Adversarial Robustnesscited by 16
  7. 2021Group Equivariant Subsamplingcited by 15
  8. 2020Shift Equivariance in Object Detectioncited by 13
  9. 2021Anti-aliasing Deep Image Classifiers using Novel Depth Adaptive Blurring and Activation Functioncited by 11
  10. 2022Generalization to translation shifts: a study in architectures and augmentationscited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 4; 2024: 0; 2025: 5; 2026: 2.

11 papers.

  1. 2023Making Vision Transformers Truly Shift-Equivariantcited by 16
  2. 2025Alias-Free Latent Diffusion Models: Improving Fractional Shift Equivariance of Diffusion Latent Spacecited by 14
  3. 2023Optimizing PatchCore for Few/many-shot Anomaly Detectioncited by 10
  4. 2025Shifting the Paradigm: A Diffeomorphism Between Time Series Data Manifolds for Achieving Shift-Invariancy in Deep Learningcited by 5
  5. 2023CoNO: Complex Neural Operator for Continuous Dynamical Systemscited by 4
  6. 2023Exponential Quantum Communication Advantage in Distributed Inference and Learningcited by 3
  7. 2025Alias-Free ViT: Fractional Shift Invariance via Linear Attentioncited by 2
  8. 2025Translation-Equivariance of Normalization Layers and Aliasing in Convolutional Neural Networkscited by 1
  9. 2025Equivariant Splitting: Self-supervised learning from incomplete datacited by 1
  10. 2026UNet-AF: Alias-free UNet architecturescited by 0
  11. 2026A Theory of Contrastive Learning with Natural Imagescited by 0