Abstract:This paper studies robust regression for data on Riemannian manifolds. Geodesic regression is the generalization of linear regression to a setting with a manifold-valued dependent variable and one or more real-valued independent variables. The existing work on geodesic regression uses the sum-of-squared errors to find the solution, but as in the classical Euclidean case, the least-squares method is highly sensitive to outliers. In this paper, we use M-type estimators, including the L1documentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$L_1$$end{document}, Huber and Tukey biweight estimators, to perform robust geodesic regression, and describe how to calculate the tuning parameters for the latter two. We show that, on compact symmetric spaces, all M-type estimators are maximum likelihood estimators, and argue in favor of a general preference for the L1documentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$L_1$$end{document} estimator over the L2documentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$L_2$$end{document} and Huber estimators on high-dimensional spaces. A derivation of the Riemannian normal distribution on Sndocumentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$S^n$$end{document} and Hndocumentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$mathbb {H}^n$$end{document} is also included. Results from numerical examples, including analysis of real neuroimaging data, demonstrate the promising empirical properties of the proposed approach.

Geodesic Discriminant Analysis on Curved Riemannian Manifold

Discriminant Analysis on Nonlinear Manifold Based on Geodesic Gabriel Graph

Geodesic Gabriel Graph Based Supervised Nonlinear Manifold Learning

A Regularized Approach for Geodesic-Based Semisupervised Multimanifold Learning

Riemannian Geodesic Discriminant Analysis–Minimum Riemannian Mean Distance: A Robust and Effective Method Leveraging a Symmetric Positive Definite Manifold and Discriminant Algorithm for Image Set Classification

Bayesian Geodesic Regression on Riemannian Manifolds

A Curvature-based Manifold Learning Algorithm.

Graph regularized linear discriminant analysis and its generalization

Riemannian Metric Learning Based on Curvature Flow.

Grassmannian Discriminant Maps (GDM) for Manifold Dimensionality Reduction with Application to Image Set Classification

Geodesic Based Semi-supervised Multi-manifold Feature Extraction

Algorithm for Riemannian Manifold Learning

GD-RDA: A New Regularized Discriminant Analysis for High-Dimensional Data

Semi-supervised discriminant analysis on Grassmannian manifold

Gaussian Differential Privacy on Riemannian Manifolds

A Modified Algorithm for Generalized Discriminant Analysis

Curvature corrected tangent space-based approximation of manifold-valued data

Curvature-aware Manifold Learning

Robust Geodesic Regression

Adaptive Neighboring Selection Algorithm Based on Curvature Prediction in Manifold Learning

Pulling back symmetric Riemannian geometry for data analysis