CNNs and Equivariance

Learning SO(3) Equivariant Representations with Spherical CNNs

It is shown that networks with much lower capacity and without requiring data augmentation can exhibit performance comparable to the state of the art in ...

10. Equivariant Neural Networks

Equivariant neural networks guarantee equivariance by construction for arbitrary groups, which removes the need to align trajectories, work in special ...

Equivariance and Symmetries in CNNs - James Urquhart Allingham

"Group equivariant convolutional networks." International conference on machine learning. 2016. Page 6. Rotations in CNNs. 1. 1. 1.

E(3)-equivariant graph neural networks for data-efficient ... - Nature

This work presents Neural Equivariant Interatomic Potentials (NequIP), an E(3)-equivariant neural network approach for learning interatomic ...

GDL - Steerable CNNs — UvA DL Notebooks v1.2 documentation

For this reason, the GCNN framework is not particularly convenient to implement networks equivariant to groups with infinite elements. Steerable CNNs are a more ...

Equivariant and Denoising CNNs to Decouple Intensity and Spatial...

We present a new registration-based motion tracking strategy that leverages equivariant and denoising CNNs to decouple intensity and spatial features.

Group Equivariant Convolutional Networks

Abstract. We introduce Group equivariant Convolutional Neural Networks (G-CNNs), a natural generalization of convolutional neural networks that reduces sample ...

Learning Steerable Filters for Rotation Equivariant CNNs

Convolutional neural networks. (CNNs) implement translational equivariance by construc- tion; for other transformations, however, they are com- pelled to ...

Reviews: General E(2)-Equivariant Steerable CNNs - NIPS papers

The paper presents extensive experimental results on a range of different architectural variants, which is informative and will be a good benchmark for future ...

Spherical CNN for Medical Imaging Applications - NCBI

... Convolutional Neural Networks (CNNs) ... To achieve equivariance, SCNN applies a group-equivariant convolutional layer that preserves rotational symmetries.

Group Equivariance and Invariance in Neural Networks

As previously mentioned, a. CNN is the most general MLP which is translation equivariant. The addition of this one constraint means that CNNs learn faster, use ...

Learning Steerable Filters for Rotation Equivariant CNNs

2 best model for Breast Tumour Classification on PCam (AUC metric)

Lie Group Equivariant Convolutional Neural Network Based on ...

We introduce a Lie group equivariant convolutional neural network predicated on the Laplace distribution.

Equivariance versus Augmentation for Spherical Images

We analyze the role of rotational equivariance in convolutional neural networks (CNNs) applied to spherical images. We compare the performance of the group ...

Andrew Gordon Wilson on X: "Translation equivariance on images ...

Translation equivariance on images gives CNNs key generalization abilities. Our new paper "Generalizing Convolutional Neural Networks for ...

CNNs on surfaces using rotation-equivariant features

Due to curvature, the transport of filter kernels on surfaces results in a rotational ambiguity, which prevents a uniform alignment of these ...

Geometric deep learning and equivariant neural networks

A gauge equivariant network for such fields consists of layers which are equivariant with respect to change of coordinates in \mathcal {M}, such ...

Efficient Equivariant Network - NIPS papers

Convolutional neural networks (CNNs) have dominated the field of Computer Vi- sion and achieved great success due to their built-in translation equivariance ...

CNNs on Surfaces using Rotation-Equivariant Features

This paper is concerned with a fundamental problem in geometric deep learning that arises in the construction of convolutional neural networks on surfaces. Due ...

3D Steerable CNNs: Learning Rotationally Equivariant Features in ...

In contrast to 3D Steerable CNNs, both networks are equivariant to certain discrete rotations only. The most closely related works achieving full SE(3) ...