O$n$ Learning Deep O($n$)-Equivariant Hyperspheres

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Melnyk, Pavlo, Felsberg, Michael, Wadenbäck, Mårten, Robinson, Andreas, Le, Cuong
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911892990066688
author Melnyk, Pavlo
Felsberg, Michael
Wadenbäck, Mårten
Robinson, Andreas
Le, Cuong
author_facet Melnyk, Pavlo
Felsberg, Michael
Wadenbäck, Mårten
Robinson, Andreas
Le, Cuong
contents In this paper, we utilize hyperspheres and regular $n$-simplexes and propose an approach to learning deep features equivariant under the transformations of $n$D reflections and rotations, encompassed by the powerful group of O$(n)$. Namely, we propose O$(n)$-equivariant neurons with spherical decision surfaces that generalize to any dimension $n$, which we call Deep Equivariant Hyperspheres. We demonstrate how to combine them in a network that directly operates on the basis of the input points and propose an invariant operator based on the relation between two points and a sphere, which as we show, turns out to be a Gram matrix. Using synthetic and real-world data in $n$D, we experimentally verify our theoretical contributions and find that our approach is superior to the competing methods for O$(n)$-equivariant benchmark datasets (classification and regression), demonstrating a favorable speed/performance trade-off. The code is available at https://github.com/pavlo-melnyk/equivariant-hyperspheres.
format Preprint
id arxiv_https___arxiv_org_abs_2305_15613
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle O$n$ Learning Deep O($n$)-Equivariant Hyperspheres
Melnyk, Pavlo
Felsberg, Michael
Wadenbäck, Mårten
Robinson, Andreas
Le, Cuong
Machine Learning
In this paper, we utilize hyperspheres and regular $n$-simplexes and propose an approach to learning deep features equivariant under the transformations of $n$D reflections and rotations, encompassed by the powerful group of O$(n)$. Namely, we propose O$(n)$-equivariant neurons with spherical decision surfaces that generalize to any dimension $n$, which we call Deep Equivariant Hyperspheres. We demonstrate how to combine them in a network that directly operates on the basis of the input points and propose an invariant operator based on the relation between two points and a sphere, which as we show, turns out to be a Gram matrix. Using synthetic and real-world data in $n$D, we experimentally verify our theoretical contributions and find that our approach is superior to the competing methods for O$(n)$-equivariant benchmark datasets (classification and regression), demonstrating a favorable speed/performance trade-off. The code is available at https://github.com/pavlo-melnyk/equivariant-hyperspheres.
title O$n$ Learning Deep O($n$)-Equivariant Hyperspheres
topic Machine Learning
url https://arxiv.org/abs/2305.15613