Saved in:
Bibliographic Details
Main Author: Nava-Yazdani, Esfandiar
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.18339
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915626228908032
author Nava-Yazdani, Esfandiar
author_facet Nava-Yazdani, Esfandiar
contents We propose a natural intrinsic extension of ridge regression from Euclidean spaces to general Riemannian manifolds for time-series prediction. Our approach combines Riemannian least-squares fitting via Bézier curves, empirical covariance on manifolds, and Mahalanobis distance regularization. A key technical contribution is an explicit formula for the gradient of the objective function using adjoint differentials, enabling efficient numerical optimization via Riemannian gradient descent. We validate our framework through synthetic spherical experiments (achieving significant error reduction over unregularized regression) and hurricane forecasting.
format Preprint
id arxiv_https___arxiv_org_abs_2411_18339
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Ridge Regression on Riemannian Manifolds for Time-Series Prediction
Nava-Yazdani, Esfandiar
Differential Geometry
Numerical Analysis
Applications
Machine Learning
53B, 53C (Primary), 62, 65D (Secondary)
We propose a natural intrinsic extension of ridge regression from Euclidean spaces to general Riemannian manifolds for time-series prediction. Our approach combines Riemannian least-squares fitting via Bézier curves, empirical covariance on manifolds, and Mahalanobis distance regularization. A key technical contribution is an explicit formula for the gradient of the objective function using adjoint differentials, enabling efficient numerical optimization via Riemannian gradient descent. We validate our framework through synthetic spherical experiments (achieving significant error reduction over unregularized regression) and hurricane forecasting.
title Ridge Regression on Riemannian Manifolds for Time-Series Prediction
topic Differential Geometry
Numerical Analysis
Applications
Machine Learning
53B, 53C (Primary), 62, 65D (Secondary)
url https://arxiv.org/abs/2411.18339