Timezone: »
We present a general theory of Group equivariant Convolutional Neural Networks (G-CNNs) on homogeneous spaces such as Euclidean space and the sphere. Feature maps in these networks represent fields on a homogeneous base space, and layers are equivariant maps between spaces of fields. The theory enables a systematic classification of all existing G-CNNs in terms of their symmetry group, base space, and field type. We also answer a fundamental question: what is the most general kind of equivariant linear map between feature spaces (fields) of given types? We show that such maps correspond one-to-one with generalized convolutions with an equivariant kernel, and characterize the space of such kernels.
Author Information
Taco Cohen (Qualcomm AI Research)
Taco Cohen is a machine learning research scientist at Qualcomm AI Research in Amsterdam and a PhD student at the University of Amsterdam, supervised by prof. Max Welling. He was a co-founder of Scyfer, a company focussed on active deep learning, acquired by Qualcomm in 2017. He holds a BSc in theoretical computer science from Utrecht University and a MSc in artificial intelligence from the University of Amsterdam (both cum laude). His research is focussed on understanding and improving deep representation learning, in particular learning of equivariant and disentangled representations, data-efficient deep learning, learning on non-Euclidean domains, and applications of group representation theory and non-commutative harmonic analysis, as well as deep learning based source compression. He has done internships at Google Deepmind (working with Geoff Hinton) and OpenAI. He received the 2014 University of Amsterdam thesis prize, a Google PhD Fellowship, ICLR 2018 best paper award for “Spherical CNNs”, and was named one of 35 innovators under 35 in Europe by MIT in 2018.
Mario Geiger (EPFL)
Maurice Weiler (University of Amsterdam)
More from the Same Authors
-
2022 : On the Expressive Power of Geometric Graph Neural Networks »
Cristian Bodnar · Chaitanya K. Joshi · Simon Mathis · Taco Cohen · Pietro Liò -
2022 : Deconfounded Imitation Learning »
Risto Vuorio · Pim de Haan · Johann Brehmer · Hanno Ackermann · Daniel Dijkman · Taco Cohen -
2023 Poster: EDGI: Equivariant Diffusion for Planning with Embodied Agents »
Johann Brehmer · Joey Bose · Pim de Haan · Taco Cohen -
2023 Poster: Geometric Algebra Transformers »
Johann Brehmer · Pim de Haan · Sönke Behrends · Taco Cohen -
2023 Poster: A General Framework for Equivariant Neural Networks on Reductive Lie Groups »
Ilyes Batatia · Mario Geiger · Jose Munoz · Tess Smidt · Lior Silberman · Christoph Ortner -
2022 : Panel Discussion I: Geometric and topological principles for representation learning in ML »
Irina Higgins · Taco Cohen · Erik Bekkers · Nina Miolane · Rose Yu -
2022 : On the Expressive Power of Geometric Graph Neural Networks »
Cristian Bodnar · Chaitanya K. Joshi · Simon Mathis · Taco Cohen · Pietro Liò -
2022 : From Equivariance to Naturality »
Taco Cohen -
2022 Poster: A PAC-Bayesian Generalization Bound for Equivariant Networks »
Arash Behboodi · Gabriele Cesa · Taco Cohen -
2022 Poster: Weakly supervised causal representation learning »
Johann Brehmer · Pim de Haan · Phillip Lippe · Taco Cohen -
2022 Poster: On the symmetries of the synchronization problem in Cryo-EM: Multi-Frequency Vector Diffusion Maps on the Projective Plane »
Gabriele Cesa · Arash Behboodi · Taco Cohen · Max Welling -
2021 Poster: SE(3)-equivariant prediction of molecular wavefunctions and electronic densities »
Oliver Unke · Mihail Bogojeski · Michael Gastegger · Mario Geiger · Tess Smidt · Klaus-Robert Müller -
2021 Poster: Relative stability toward diffeomorphisms indicates performance in deep nets »
Leonardo Petrini · Alessandro Favero · Mario Geiger · Matthieu Wyart -
2020 Poster: Natural Graph Networks »
Pim de Haan · Taco Cohen · Max Welling -
2020 Tutorial: (Track2) Equivariant Networks Q&A »
Risi Kondor · Taco Cohen -
2020 Tutorial: (Track2) Equivariant Networks »
Risi Kondor · Taco Cohen -
2019 Poster: General E(2)-Equivariant Steerable CNNs »
Maurice Weiler · Gabriele Cesa -
2018 Poster: 3D Steerable CNNs: Learning Rotationally Equivariant Features in Volumetric Data »
Maurice Weiler · Wouter Boomsma · Mario Geiger · Max Welling · Taco Cohen