Learnable Polyphase Sampling for Shift Invariant and Equivariant Convolutional Networks

Abstract

We propose learnable polyphase sampling (LPS), a pair of learnable down/upsampling layers that enable truly shift-invariant and equivariant convolutional networks. LPS can be trained end-to-end from data and generalizes existing handcrafted downsampling layers. It is widely applicable as it can be integrated into any convolutional network by replacing down/upsampling layers. We evaluate LPS on image classification and semantic segmentation. Experiments show that LPS is on-par with or outperforms existing methods in both performance and shift consistency. For the first time, we achieve true shift-equivariance on semantic segmentation (PASCAL VOC), i.e., 100% shift consistency, outperforming baselines by an absolute 3.3%.

PDF NeurIPS OpenReview Semantic Scholar

Cite

Text

Rojas-Gomez et al. "Learnable Polyphase Sampling for Shift Invariant and Equivariant Convolutional Networks." Neural Information Processing Systems, 2022.

Markdown

[Rojas-Gomez et al. "Learnable Polyphase Sampling for Shift Invariant and Equivariant Convolutional Networks." Neural Information Processing Systems, 2022.](https://mlanthology.org/neurips/2022/rojasgomez2022neurips-learnable/)

BibTeX

@inproceedings{rojasgomez2022neurips-learnable,
  title     = {{Learnable Polyphase Sampling for Shift Invariant and Equivariant Convolutional Networks}},
  author    = {Rojas-Gomez, Renan A. and Lim, Teck-Yian and Schwing, Alex and Do, Minh and Yeh, Raymond A.},
  booktitle = {Neural Information Processing Systems},
  year      = {2022},
  url       = {https://mlanthology.org/neurips/2022/rojasgomez2022neurips-learnable/}
}