SyncAI.news, a Varaisys broadcasting
Graded Representation Theory of Equivariant Neural Networks
MS

Mani Shayestehfar

· 1 min read

ResearcharXiv cs.LG

Graded Representation Theory of Equivariant Neural Networks

arXiv:2609.25776v1 Announce Type: cross Abstract: Nonlinear activations can create equivariant interactions between irreducible representations that linear maps cannot. We use the Gaussian degree decomposition to extend ordinary polynomial degree to such nonlinear maps, and prove that for a fixed coordinatewise equivariant layer each degree factors into a polynomial determined by the linear maps and a scalar determined by the activation. This separates three distinct obstructions, coming from symmetry, coordinates, and activation.

Original source

This story was published by arXiv cs.LG and written by Mani Shayestehfar. SyncAI.news shows a preview; the complete article is on the publisher's site.

Read the full story on arxiv.org

Similar News