The decimation scheme for symmetric matrix factorization

F Camilli, M Mézard - Journal of Physics A: Mathematical and …, 2024 - iopscience.iop.org
Matrix factorization is an inference problem that has acquired importance due to its vast
range of applications that go from dictionary learning to recommendation systems and …

Matrix inference in growing rank regimes

F Pourkamali, J Barbier, N Macris - IEEE Transactions on …, 2024 - ieeexplore.ieee.org
The inference of a large symmetric signal-matrix S∈ RN× N corrupted by additive Gaussian
noise, is considered for two regimes of growth of the rank M as a function of N. For sub-linear …

Rectangular rotational invariant estimator for general additive noise matrices

F Pourkamali, N Macris - 2023 IEEE International Symposium …, 2023 - ieeexplore.ieee.org
We propose a rectangular rotational invariant estimator to recover a real matrix from noisy
matrix observations coming from an arbitrary additive rotational invariant perturbation, in the …

Bayesian extensive-rank matrix factorization with rotational invariant priors

F Pourkamali, N Macris - Advances in Neural Information …, 2024 - proceedings.neurips.cc
We consider a statistical model for matrix factorization in a regime where the rank of the two
hidden matrix factors grows linearly with their dimension and their product is corrupted by …

Matrix denoising: Bayes-optimal estimators via low-degree polynomials

G Semerjian - arXiv preprint arXiv:2402.16719, 2024 - arxiv.org
We consider the additive version of the matrix denoising problem, where a random
symmetric matrix $ S $ of size $ n $ has to be inferred from the observation of $ Y= S+ Z …

Gradient flow on extensive-rank positive semi-definite matrix denoising

A Bodin, N Macris - 2023 IEEE Information Theory Workshop …, 2023 - ieeexplore.ieee.org
In this work, we present a new approach to analyze the gradient flow for a positive semi-
definite matrix denoising problem in an extensive-rank and high-dimensional regime. We …

Under-parameterized double descent for ridge regularized least squares denoising of data on a line

R Sonthalia, X Li, B Gu - arXiv preprint arXiv:2305.14689, 2023 - arxiv.org
The relationship between the number of training data points, the number of parameters in a
statistical model, and the generalization capabilities of the model has been widely studied …

Spherical integrals of sublinear rank

J Husson, J Ko - arXiv preprint arXiv:2208.03642, 2022 - arxiv.org
We consider the asymptotics of $ k $-dimensional spherical integrals when $ k= o (N) $. We
prove that the $ o (N) $-dimensional spherical integrals are approximately the products of $1 …

Singular vectors of sums of rectangular random matrices and optimal estimation of high-rank signals: The extensive spike model

ID Landau, GC Mel, S Ganguli - Physical Review E, 2023 - APS
Across many disciplines spanning from neuroscience and genomics to machine learning,
atmospheric science, and finance, the problems of denoising large data matrices to recover …

Fitting an ellipsoid to random points: predictions using the replica method

A Maillard, D Kunisky - IEEE Transactions on Information …, 2024 - ieeexplore.ieee.org
We consider the problem of fitting a centered ellipsoid to n standard Gaussian random
vectors in R d, as n, d→∞ with n/d 2→ α> 0. It has been conjectured that this problem is, with …