Attentive Group Equivariant Convolutional Networks

By David W. Romero, Erik J. Bekkers, J. Tomczak, Mark Hoogendoorn. 2020

102 citations recorded by the source corpus.

Although group convolutional networks are able to learn powerful representations based on symmetry patterns, they lack explicit means to learn meaningful relationships among them (e.g., relative positions and poses). In this paper, we present attentive group equivariant convolutions, a generalization of the group convolution, in which attention is applied during the course of convolution to accentuate meaningful symmetry combinations and suppress non-plausible, misleading ones. We indicate that prior work on visual attention can be described as special cases of our proposed framework and show empirically that our attentive group equivariant convolutional networks consistently outperform conventional group convolutional networks on benchmark image datasets. Simultaneously, we provide interpretability to the learned concepts through the visualization of equivariant attention maps.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Group Equivariant Stand-Alone Self-Attention For Vision (2020, background): “More recently, attention has been introduced to group equivariant architectures (Romero & Hoogendoorn, 2019; Diaconu & Worrall, 2019; Romero et al., 2020a).”
SE(3)-Transformers: 3D Roto-Translation Equivariant Attention Networks (2020, methodology): “[24] recently introduced a 2D roto-translationally equivariant attention module for pixel-based image data.”
Wavelet Networks: Scale Equivariant Learning From Raw Waveforms (2020, methodology): “Furthermore, we would like to replace group convolutions by their attentive counterparts [55, 56] to construct spectrotemporal attention maps indicating the importance of particular time-frequency atoms for a given task.”
Geometric and Physical Quantities improve E(3) Equivariant Message Passing (2021, background): “…written in convolution form where, in the latter case, the linear transformations additionally depend on an input dependent attention mechanism as in (Anderson et al., 2019; Fuchs et al., 2020), and can be seen as a steerable PointConv version of attentive group convolutions (Romero et al., 2020).”
Probabilistic Symmetries and Invariant Neural Networks (2020, background): “Modifications of the basic convolutional filters have appeared in the form of convolutional capsules (e.g., Sabour et al., 2017; Hinton et al., 2018; Lenssen et al., 2018), and convolution was recently combined with attention mechanisms (Bello et al., 2019; Romero et al., 2020).”
Attention mechanism in neural networks: where it comes and where it goes (2022, background): “Another interesting study introduces attentive group convolutions with a generalization of visual self-attention [96].”