Parallel Transport on the Cone Manifold of SPD Matrices for Domain Adaptation

Abstract

In this paper, we consider the problem of domain adaptation. We propose to view the data through the lens of covariance matrices and present a method for domain adaptation using parallel transport on the cone manifold of symmetric positive-definite matrices. We provide rigorous analysis using Riemanninan geometry, illuminating the theoretical guarantees and benefits of the presented method. In addition, we demonstrate these benefits using experimental results on simulations and real-measured data.

References in corpus (4)

• Kernel methods in machine learning
• Conic geometric optimisation on the manifold of positive definite matrices
• Optimal Bayesian Transfer Learning
• Priors with Coupled First and Second Order Differences for Manifold-Valued Image Processing

Cited by in corpus (12)

• Geometry-aware Dynamic Movement Primitives
• MIG Median Detectors with Manifold Filter
• Unsupervised Learning Discriminative MIG Detectors in Nonhomogeneous Clutter
• Transfer Learning for EEG-Based Brain-Computer Interfaces: A Review of Progress Made Since 2016
• Revisiting Euclidean Alignment for Transfer Learning in EEG-Based Brain-Computer Interfaces
• Separable Shape Tensors for Aerodynamic Design
• Manifold Learning for Organizing Unstructured Sets of Process Observations
• A reduced parallel transport equation on Lie Groups with a left-invariant metric
• Domain Adaptation with Optimal Transport on the Manifold of SPD matrices
• Deep Optimal Transport for Domain Adaptation on SPD Manifolds
• Spectral Flow on the Manifold of SPD Matrices for Multimodal Data Processing
• GeomNet: A Neural Network Based on Riemannian Geometries of SPD Matrix Space and Cholesky Space for 3D Skeleton-Based Interaction Recognition