Fundamental limits of low-rank matrix estimation: the non-symmetric case

Abstract

We consider the high-dimensional inference problem where the signal is a low-rank matrix which is corrupted by an additive Gaussian noise. Given a probabilistic model for the low-rank matrix, we compute the limit in the large dimension setting for the mutual information between the signal and the observations, as well as the matrix minimum mean square error, while the rank of the signal remains constant. This allows to locate the information-theoretic threshold for this estimation problem, i.e. the critical value of the signal intensity below which it is impossible to recover the low-rank matrix.

References in corpus (1)

• Optimality and Sub-optimality of PCA for Spiked Random Matrices and Synchronization

Cited by in corpus (13)

• Bilinear Recovery using Adaptive Vector-AMP
• Statistical limits of dictionary learning: random matrix theory and the spectral replica method
• Subexponential-Time Algorithms for Sparse PCA
• 0-1 phase transitions in sparse spiked matrix estimation
• Detection limits in the high-dimensional spiked rectangular model
• Adapting to Unknown Noise Distribution in Matrix Denoising
• Thresholds of descending algorithms in inference problems
• Weak Detection in the Spiked Wigner Model with General Rank
• Community Detection with Contextual Multilayer Networks
• Efficient inference in stochastic block models with vertex labels
• Neural-prior stochastic block model
• Empirical Bayes PCA in high dimensions
• TAP free energy, spin glasses, and variational inference