Here, the functionf() is applied toAx in a component-wise manner. The above model arises in many applications of signal processing [13, 10, 41], communications [56, 9, 25], and machine learning [48, 40].

Similarly the ISOMAP face database consists ofimages (256levels ofgray)ofsize64 64,i.e.,vectors in R4096, whereas the correct intrinsic dimension is only3 (for the vertical, horizontal pause and lightingdirection). The second approach, is anaverage caseapproach (in the spirit of thestatistical mechanics treatment ofhighdimensional systems), thatmodelsfeaturevectorsby arandom ensemble,taken as aset ofrandom vectors with independently identically distributed (i.i.d.) components, and a small but xed fraction of non-zero components.

The problem of learning the parameters of a neural network is two-fold. First, we want that their training on a set of data via minimization of a suitable loss function succeed in finding a set of parameters for which the value of the loss is close to its global minimum.

V AMP's key attraction is that it applies to a larger class of matrices A than standard AMP

AMP to get insights into the statistical properties of SLOPE.

For generative priors with fully connected layers and Gaussian i.i.d.

Q2: Please summarize your review in 1-2 sentences very nice method, important application, could benefit from more explicit comparisons and demonstration of scalability.

Neural scaling laws underlie many of the recent advances in deep learning, yet their theoretical understanding remains largely confined to linear models. In this work, we present a systematic analysis of scaling laws for quadratic and diagonal neural networks in the feature learning regime. Leveraging connections with matrix compressed sensing and LASSO, we derive a detailed phase diagram for the scaling exponents of the excess risk as a function of sample complexity and weight decay. This analysis uncovers crossovers between distinct scaling regimes and plateau behaviors, mirroring phenomena widely reported in the empirical neural scaling literature. Furthermore, we establish a precise link between these regimes and the spectral properties of the trained network weights, which we characterize in detail. As a consequence, we provide a theoretical validation of recent empirical observations connecting the emergence of power-law tails in the weight spectrum with network generalization performance, yielding an interpretation from first principles.

We establish non-asymptotic bounds on the finite-sample behavior of generalized first-order iterative algorithms -- including gradient-based optimization methods and approximate message passing (AMP) -- with Gaussian data matrices and full-memory, non-separable nonlinearities. The central result constructs an explicit coupling between the iterates of a generalized first-order method and a conditionally Gaussian process whose covariance evolves deterministically via a finite-dimensional state evolution recursion. This coupling yields tight, dimension-free bounds under mild Lipschitz and moment-matching conditions. Our analysis departs from classical inductive AMP proofs by employing a direct comparison between the generalized first-order method and the conditionally Gaussian comparison process. This approach provides a unified derivation of AMP theory for Gaussian matrices without relying on separability or asymptotics. A complementary lower bound on the Wasserstein distance demonstrates the sharpness of our upper bounds.