Unit-Consistent (UC) Adjoint for GSD and Backprop in Deep Learning Applications

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autor principal: Uhlmann, Jeffrey
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918291781451776
author Uhlmann, Jeffrey
author_facet Uhlmann, Jeffrey
contents Deep neural networks constructed from linear maps and positively homogeneous nonlinearities (e.g., ReLU) possess a fundamental gauge symmetry: the network function is invariant to node-wise diagonal rescalings. However, standard gradient descent is not equivariant to this symmetry, causing optimization trajectories to depend heavily on arbitrary parameterizations. Prior work has proposed rescaling-invariant optimization schemes for positively homogeneous networks (e.g., path-based or path-space updates). Our contribution is complementary: we formulate the invariance requirement at the level of the backward adjoint/optimization geometry, which provides a simple, operator-level recipe that can be applied uniformly across network components and optimizer state. By replacing the Euclidean transpose with a Unit-Consistent (UC) adjoint, we derive UC gauge-consistent steepest descent and backprogation.
format Preprint
id arxiv_https___arxiv_org_abs_2601_10873
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Unit-Consistent (UC) Adjoint for GSD and Backprop in Deep Learning Applications
Uhlmann, Jeffrey
Machine Learning
Deep neural networks constructed from linear maps and positively homogeneous nonlinearities (e.g., ReLU) possess a fundamental gauge symmetry: the network function is invariant to node-wise diagonal rescalings. However, standard gradient descent is not equivariant to this symmetry, causing optimization trajectories to depend heavily on arbitrary parameterizations. Prior work has proposed rescaling-invariant optimization schemes for positively homogeneous networks (e.g., path-based or path-space updates). Our contribution is complementary: we formulate the invariance requirement at the level of the backward adjoint/optimization geometry, which provides a simple, operator-level recipe that can be applied uniformly across network components and optimizer state. By replacing the Euclidean transpose with a Unit-Consistent (UC) adjoint, we derive UC gauge-consistent steepest descent and backprogation.
title Unit-Consistent (UC) Adjoint for GSD and Backprop in Deep Learning Applications
topic Machine Learning
url https://arxiv.org/abs/2601.10873