Equivariant Manifold Flows
Abstract
Tractably modelling distributions over manifolds has long been an important goal in the natural sciences. Recent work has focused on developing general machine learning models to learn such distributions. However, for many applications these distributions must respect manifold symmetries—a trait which most previous models disregard. In this paper, we lay the theoretical foundations for learning symmetry-invariant distributions on arbitrary manifolds via equivariant manifold flows. We demonstrate the utility of our approach by learning quantum field theory-motivated invariant SU(n) densities and by correcting meteor impact dataset bias.
Cite
Text
Katsman et al. "Equivariant Manifold Flows." Neural Information Processing Systems, 2021.Markdown
[Katsman et al. "Equivariant Manifold Flows." Neural Information Processing Systems, 2021.](https://mlanthology.org/neurips/2021/katsman2021neurips-equivariant/)BibTeX
@inproceedings{katsman2021neurips-equivariant,
title = {{Equivariant Manifold Flows}},
author = {Katsman, Isay and Lou, Aaron and Lim, Derek and Jiang, Qingxuan and Lim, Ser Nam and De Sa, Christopher M},
booktitle = {Neural Information Processing Systems},
year = {2021},
url = {https://mlanthology.org/neurips/2021/katsman2021neurips-equivariant/}
}