Guardado en:
Detalles Bibliográficos
Autores principales: Lin, Wu, Duruisseaux, Valentin, Leok, Melvin, Nielsen, Frank, Khan, Mohammad Emtiyaz, Schmidt, Mark
Formato: Preprint
Publicado: 2023
Materias:
Acceso en línea:https://arxiv.org/abs/2302.09738
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909138636767232
author Lin, Wu
Duruisseaux, Valentin
Leok, Melvin
Nielsen, Frank
Khan, Mohammad Emtiyaz
Schmidt, Mark
author_facet Lin, Wu
Duruisseaux, Valentin
Leok, Melvin
Nielsen, Frank
Khan, Mohammad Emtiyaz
Schmidt, Mark
contents Riemannian submanifold optimization with momentum is computationally challenging because, to ensure that the iterates remain on the submanifold, we often need to solve difficult differential equations. Here, we simplify such difficulties for a class of sparse or structured symmetric positive-definite matrices with the affine-invariant metric. We do so by proposing a generalized version of the Riemannian normal coordinates that dynamically orthonormalizes the metric and locally converts the problem into an unconstrained problem in the Euclidean space. We use our approach to simplify existing approaches for structured covariances and develop matrix-inverse-free $2^\text{nd}$-order optimizers for deep learning with low precision by using only matrix multiplications. Code: https://github.com/yorkerlin/StructuredNGD-DL
format Preprint
id arxiv_https___arxiv_org_abs_2302_09738
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Simplifying Momentum-based Positive-definite Submanifold Optimization with Applications to Deep Learning
Lin, Wu
Duruisseaux, Valentin
Leok, Melvin
Nielsen, Frank
Khan, Mohammad Emtiyaz
Schmidt, Mark
Machine Learning
Riemannian submanifold optimization with momentum is computationally challenging because, to ensure that the iterates remain on the submanifold, we often need to solve difficult differential equations. Here, we simplify such difficulties for a class of sparse or structured symmetric positive-definite matrices with the affine-invariant metric. We do so by proposing a generalized version of the Riemannian normal coordinates that dynamically orthonormalizes the metric and locally converts the problem into an unconstrained problem in the Euclidean space. We use our approach to simplify existing approaches for structured covariances and develop matrix-inverse-free $2^\text{nd}$-order optimizers for deep learning with low precision by using only matrix multiplications. Code: https://github.com/yorkerlin/StructuredNGD-DL
title Simplifying Momentum-based Positive-definite Submanifold Optimization with Applications to Deep Learning
topic Machine Learning
url https://arxiv.org/abs/2302.09738