Skip to content
Preprint

Orthogonal Polynomial Approximation for Matrix Log Normalization in Global Covariance Pooling

Aug 2026 · 0 citations · 38 references
Computer Science

TL;DR

The decomposition-free logarithm is both faster and more accurate than the spectral logarithm and than the square-root approximations it replaces, and at matched basis and degree the log target beats the square-root target, confirming that the gain comes from the faithful Riemannian map rather than from a better polynomial family.

Abstract

Global Covariance Pooling (GCP) improves deep networks by capturing second-order feature statistics, and is especially effective for fine-grained recognition. Because covariance matrices live on the Symmetric Positive Definite (SPD) manifold, a normalization step is required before the Euclidean classifier. The faithful choice is the matrix logarithm (MLN-COV), which maps the SPD manifold to its tangent space; in practice it was abandoned in favour of the matrix square root because its eigendecomposition-based gradient is numerically unstable. We show that this instability is an artifact of computing the logarithm spectrally, not of the logarithm itself. Approximating the logarithm with finite polynomials in the covariance matrix removes the eigendecomposition from both passes: every operation becomes a General Matrix Multiplication (GEMM), the gradient stays bounded on the spectral support of the pre-normalized covariance, and the unstable 1/(lambda_i-lambda_j) term never appears. The key ingredient is a mean-eigenvalue pre-normalization that centres the spectrum near 1, away from the singularity of log, with a scalar post-compensation that returns the singular part of log(A) in closed form. Our recommended normalizer is a degree-8 Chebyshev expansion evaluated by a three-term matrix recurrence, with a matching reverse recurrence for the backward pass; Legendre, Laguerre, Taylor and Pade expansions are studied as controls that isolate the roles of the basis and of the target function. On three fine-grained benchmarks and ImageNet-1k the decomposition-free logarithm is both faster and more accurate than the spectral logarithm and than the square-root approximations it replaces, and at matched basis and degree the log target beats the square-root target, confirming that the gain comes from the faithful Riemannian map rather than from a better polynomial family.

View source

Similar papers

Preprint Oct 2026

A Balanced Parametrization of Correlation Matrices

We introduce the balanced correlation parametrization (BCP), which maps positive-definite correlation matrices one-to-one onto the set of symmetric hollow (zero-diagonal) matrices with unrestricted off-diagonal entries. A key feature of the parametrization is that both the transformation and its inverse are available i...

P. Hansen, Tong Chen · 0 citations
Preprint Oct 2026

First-order Optimization Methods with Generalized Smoothness for Low-rank Matrix Factorization

This paper studies several first-order methods for low-rank matrix minimization in the Burer--Monteiro form, where the target matrix is represented as the product of a nonnegative factor matrix and its transpose. Such a factorization destroys convexity of the objective function and global Lipschitz continuity of the gr...

Александр Владимирович Биданец, F. Stonyakin · 0 citations
Preprint Sep 2026

Riemannian Gradient Descent for Gaussian Mixture Models with unknown diagonal covariances

This paper investigates the numerical resolution of the Beurling-LASSO (BLASSO), a convex optimization framework that promotes sparsity in the space of measures. We consider its application to the estimation of Gaussian mixture models (GMMs) with an unknown number of components and unknown diagonal covariance matrices....

Romane Giard, Y. de Castro, Roland Denis et al. · 0 citations
#machine learning Preprint Sep 2026

Understanding the Subspace Stabilization of the Hessian and Gradient Covariance Matrix

In the gradient flow set-up, it is shown that the slow evolution of the idenfied approximation is due to the separation between the outlier and the bulk eigenvalues of the Hessian matrix, thus providing an explanation to the phenomenon of the top subspace stabilization of the Hessian matrix.

Fang-Shuo Liao, Anastasios Kyrillidis · 0 citations
#machine learning Preprint Oct 2026

SCORE: Spectral Correlation Estimation for Multivariate Gaussians

Neural network-based predictive modeling with high-dimensional structured Gaussian targets requires an efficient and numerically stable, yet expressive approximation of the covariance matrix. We propose SCORE: a scalable framework, combining scoring rule training with an expressive covariance approximation learned in s...

Christopher Bulte, Emil Partow, Astha Gupta et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.