Geometric statistics with subspace structure preservation for SPD matrices

2024-07-02 · Cyrus Mostajeran, Nathaël Da Costa, Graham Van Goffrier, Rodolphe Sepulchre

We present a geometric framework for the processing of SPD-valued data that preserves subspace structures and is based on the efficient computation of extreme generalized eigenvalues. This is achieved through the use of the Thompson geometry of the semidefinite cone. We explore a particular geodesic space structure in detail and establish several properties associated with it. Finally, we review a novel inductive mean of SPD matrices based on this geometry.

📄 PDF Abstract BibTeX arXiv:2407.03382

Code (0)

등록된 구현이 없습니다.

Geometric structures and deviations on James' symmetric positive-definite matrix bicone domain

2026-03-03 · Jacek Karwowski, Frank Nielsen arxiv

Symmetric positive-definite (SPD) matrix datasets play a central role across numerous scientific disciplines, including signal processing, statistics, finance, computer vision, information theory, and machine learning am…

Hypothesis testing on invariant subspaces of non-symmetric matrices with applications to network statistics

2023-03-31 · Jérôme R. Simons

We extend the inference procedure for eigenvectors of Tyler (1981), which assumes symmetrizable matrices to generic invariant and singular subspaces of non-diagonalisable matrices to test whether $\nu \in \mathbb{R}^{p \…

Geometric optimisation on positive definite matrices for elliptically contoured distributions

2013-12-01 · NeurIPS 2013 12 · Suvrit Sra, Reshad Hosseini

Hermitian positive definite matrices (HPD) recur throughout statistics and machine learning. In this paper we develop \emph{geometric optimisation} for globally optimising certain nonconvex loss functions arising in the …

Riemannian optimization

Flag Varieties: A Geometric Framework for Deep Network Alignment

2026-05-11 · Jingchuan Xiao, Xinyi Sui, Cihan Ruan arxiv

Alignment, the tendency of adjacent weight matrices in deep networks to develop compatible subspace orientations, underlies gradient flow, Neural Collapse, and representation similarity across architectures. Despite exte…

CASC: Causal Adversarial Subspace Clustering for Multivariate Spatiotemporal Data

2026-07-23 · Francis Ndikum Nji, Vandana Janeja, Jianwu Wang arxiv

Deep subspace clustering plays a critical role in applications involving multivariate spatiotemporal data, such as sea ice monitoring, disease spread analysis, and tracking neuro-degeneration over time. Despite recent ad…