Equivariant neural networks and piecewise linear representation theory

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Gibson, Joel, Tubbenhauer, Daniel, Williamson, Geordie
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915893672411136
author Gibson, Joel
Tubbenhauer, Daniel
Williamson, Geordie
author_facet Gibson, Joel
Tubbenhauer, Daniel
Williamson, Geordie
contents Equivariant neural networks are neural networks with symmetry. Motivated by the theory of group representations, we decompose the layers of an equivariant neural network into simple representations. The nonlinear activation functions lead to interesting nonlinear equivariant maps between simple representations. For example, the rectified linear unit (ReLU) gives rise to piecewise linear maps. We show that these considerations lead to a filtration of equivariant neural networks, generalizing Fourier series. This observation might provide a useful tool for interpreting equivariant neural networks.
format Preprint
id arxiv_https___arxiv_org_abs_2408_00949
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Equivariant neural networks and piecewise linear representation theory
Gibson, Joel
Tubbenhauer, Daniel
Williamson, Geordie
Machine Learning
Group Theory
Representation Theory
Primary: 20C05, Secondary: 05E10, 68T07
Equivariant neural networks are neural networks with symmetry. Motivated by the theory of group representations, we decompose the layers of an equivariant neural network into simple representations. The nonlinear activation functions lead to interesting nonlinear equivariant maps between simple representations. For example, the rectified linear unit (ReLU) gives rise to piecewise linear maps. We show that these considerations lead to a filtration of equivariant neural networks, generalizing Fourier series. This observation might provide a useful tool for interpreting equivariant neural networks.
title Equivariant neural networks and piecewise linear representation theory
topic Machine Learning
Group Theory
Representation Theory
Primary: 20C05, Secondary: 05E10, 68T07
url https://arxiv.org/abs/2408.00949