Pith. sign in

REVIEW 6 cited by

A Random Matrix Theory Perspective on the Spectrum of Learned Features and Asymptotic Generalization Capabilities

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.18938 v1 pith:M6H5WMID submitted 2024-10-24 stat.ML cs.LGmath.STstat.TH

A Random Matrix Theory Perspective on the Spectrum of Learned Features and Asymptotic Generalization Capabilities

classification stat.ML cs.LGmath.STstat.TH
keywords featurerandomfeaturesgeneralizationlearningmatrixtrainingasymptotic
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

A key property of neural networks is their capacity of adapting to data during training. Yet, our current mathematical understanding of feature learning and its relationship to generalization remain limited. In this work, we provide a random matrix analysis of how fully-connected two-layer neural networks adapt to the target function after a single, but aggressive, gradient descent step. We rigorously establish the equivalence between the updated features and an isotropic spiked random feature model, in the limit of large batch size. For the latter model, we derive a deterministic equivalent description of the feature empirical covariance matrix in terms of certain low-dimensional operators. This allows us to sharply characterize the impact of training in the asymptotic feature spectrum, and in particular, provides a theoretical grounding for how the tails of the feature spectrum modify with training. The deterministic equivalent further yields the exact asymptotic generalization error, shedding light on the mechanisms behind its improvement in the presence of feature learning. Our result goes beyond standard random matrix ensembles, and therefore we believe it is of independent technical interest. Different from previous work, our result holds in the challenging maximal learning rate regime, is fully rigorous and allows for finitely supported second layer initialization, which turns out to be crucial for studying the functional expressivity of the learned features. This provides a sharp description of the impact of feature learning in the generalization of two-layer neural networks, beyond the random features and lazy training regimes.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Random Matrix Spectra from Boltzmann-Weighted Lattice Ensembles

    cond-mat.dis-nn 2026-05 unverdicted novelty 7.0

    A framework maps Boltzmann-weighted lattice configurations to correlated random matrix ensembles via real-space to momentum-space variance profiles, deriving spectral moments and resolvent densities benchmarked on Isi...

  2. Scaling Laws from Sequential Feature Recovery: A Solvable Hierarchical Model

    stat.ML 2026-05 accept novelty 7.0

    A solvable hierarchical model with power-law feature strengths yields explicit power-law scaling of prediction error through sequential recovery of latent directions by a layer-wise spectral algorithm.

  3. Spectral Dynamics in Deep Networks: Feature Learning, Outlier Escape, and Learning Rate Transfer

    cond-mat.dis-nn 2026-05 unverdicted novelty 7.0

    A two-level DMFT predicts width-consistent outlier escape and hyperparameter transfer under μP in deep networks, with bulk restructuring dominating for tasks with many outputs.

  4. Spectral phase transitions and trainability in neural network learning dynamics

    cond-mat.dis-nn 2026-06 unverdicted novelty 6.0

    SGD on neural network weights induces a BBP phase transition that detaches signal eigenvalues from the random bulk, yielding an analytically solvable phase diagram for trainability in a linear teacher-student model.

  5. Spectral Dynamics in Deep Networks: Feature Learning, Outlier Escape, and Learning Rate Transfer

    cond-mat.dis-nn 2026-05 unverdicted novelty 6.0

    A two-level DMFT tracks bulk and outlier spectral dynamics in wide networks, predicting width-consistent outlier growth and hyperparameter transfer under muP scaling for deep linear nets while noting bulk restructurin...

  6. HTMuon: Improving Muon via Heavy-Tailed Spectral Correction

    cs.LG 2026-03 unverdicted novelty 5.0

    HTMuon modifies Muon to produce heavier-tailed updates and weight spectra via HT-SR theory, yielding up to 0.98 lower perplexity on LLaMA pretraining and serving as a plug-in for other Muon variants.