Id: arxiv/2602.09240 Url: https://arxiv.org/abs/2602.09240 Tags: math.ST, cs.IT, cs.LG, math.PR, stat.ML Score: 0 Title: Optimal Estimation in Orthogonally Invariant Generalized Linear Models: Spectral Initialization and Approximate Message Passing Domain: math.ST Source: arxiv Authors: Yihan Zhang, Hong Chang Ji, Ramji Venkataramanan, Marco Mondelli Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-02-09 Page: https://delphi.science/r/arxiv/2602.09240 We consider the problem of parameter estimation from a generalized linear model with a random design matrix that is orthogonally invariant in law. Such a model allows the design have an arbitrary distribution of singular values and only assumes that its singular vectors are generic. It is a vast generalization of the i.i.d. Gaussian design typically considered in the theoretical literature, and is motivated by the fact that real data often have a complex correlation structure so that methods relying on i.i.d. assumptions can be highly suboptimal. Building on the paradigm of…