Inversion-Free Natural Gradient Descent on Riemannian Manifolds

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Draca, Dario, Matsubara, Takuo, Tran, Minh-Ngoc
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917546149543936
author Draca, Dario
Matsubara, Takuo
Tran, Minh-Ngoc
author_facet Draca, Dario
Matsubara, Takuo
Tran, Minh-Ngoc
contents The natural gradient method is a central tool for statistical optimisation, but its broader application is hindered by the assumption of a Euclidean parameter space, the repeated estimation of the Fisher information matrix (FIM), and the computational cost of its subsequent inversion. This paper proposes an intrinsic, inversion-free natural gradient method for statistical models whose parameters lie on general Riemannian manifolds. Formulating statistical optimisation in this non-Euclidean setting allows for the natural enforcement of parameter constraints, the elimination of non-identifiable parameters, and the exploitation of geodesic convexity. Our algorithm is based on a moving approximation of the inverse FIM, which is maintained directly on the manifold. This approximation is efficiently updated with new score vectors using low-rank matrix identities. We prove almost-sure convergence rates of $O(\log s / s^α)$ for the sequence of iterates, and a similar rate for the approximate FIM. A limited-memory variant with sub-quadratic storage complexity is further proposed for large-scale applications. We demonstrate the efficacy of our method on variational Bayes within the Bures-Wasserstein manifold, normalising flows on the Stiefel manifold, and reduced-rank logistic regression.
format Preprint
id arxiv_https___arxiv_org_abs_2604_02969
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Inversion-Free Natural Gradient Descent on Riemannian Manifolds
Draca, Dario
Matsubara, Takuo
Tran, Minh-Ngoc
Machine Learning
Computation
Methodology
The natural gradient method is a central tool for statistical optimisation, but its broader application is hindered by the assumption of a Euclidean parameter space, the repeated estimation of the Fisher information matrix (FIM), and the computational cost of its subsequent inversion. This paper proposes an intrinsic, inversion-free natural gradient method for statistical models whose parameters lie on general Riemannian manifolds. Formulating statistical optimisation in this non-Euclidean setting allows for the natural enforcement of parameter constraints, the elimination of non-identifiable parameters, and the exploitation of geodesic convexity. Our algorithm is based on a moving approximation of the inverse FIM, which is maintained directly on the manifold. This approximation is efficiently updated with new score vectors using low-rank matrix identities. We prove almost-sure convergence rates of $O(\log s / s^α)$ for the sequence of iterates, and a similar rate for the approximate FIM. A limited-memory variant with sub-quadratic storage complexity is further proposed for large-scale applications. We demonstrate the efficacy of our method on variational Bayes within the Bures-Wasserstein manifold, normalising flows on the Stiefel manifold, and reduced-rank logistic regression.
title Inversion-Free Natural Gradient Descent on Riemannian Manifolds
topic Machine Learning
Computation
Methodology
url https://arxiv.org/abs/2604.02969