Pith. sign in

REVIEW 3 cited by

High-dimensional manifold of solutions in neural networks: insights from statistical physics

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 2309.09240 v2 pith:UYPAE5VE submitted 2023-09-17 cond-mat.dis-nn cs.LGmath.PRmath.STstat.TH

High-dimensional manifold of solutions in neural networks: insights from statistical physics

classification cond-mat.dis-nn cs.LGmath.PRmath.STstat.TH
keywords solutionsolutionsapproachbinarygiveinsightsmanifoldnetworks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In these pedagogic notes I review the statistical mechanics approach to neural networks, focusing on the paradigmatic example of the perceptron architecture with binary an continuous weights, in the classification setting. I will review the Gardner's approach based on replica method and the derivation of the SAT/UNSAT transition in the storage setting. Then, I discuss some recent works that unveiled how the zero training error configurations are geometrically arranged, and how this arrangement changes as the size of the training set increases. I also illustrate how different regions of solution space can be explored analytically and how the landscape in the vicinity of a solution can be characterized. I give evidence how, in binary weight models, algorithmic hardness is a consequence of the disappearance of a clustered region of solutions that extends to very large distances. Finally, I demonstrate how the study of linear mode connectivity between solutions can give insights into the average shape of the solution manifold.

discussion (0)

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

Forward citations

Cited by 3 Pith papers

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

  1. On the robustness of noisy solutions in non-convex neural networks

    cond-mat.dis-nn 2026-07 conditional novelty 6.0

    Finite training error extends the overlap-gap threshold in binary perceptrons so wide, algorithmically reachable basins persist and still generalize where zero-error solutions are hard.

  2. Shortcomings and capacities of real-constrained neural networks in complex spaces

    cs.LG 2026-06 unverdicted novelty 6.0

    Derives the asymptotic ratio of storage capacities between real-constrained and complex pre-activations in complex neural networks using Gardner volumes and the HCIZ formula.

  3. Pseudo quantum advantages in perceptron storage capacity

    quant-ph 2025-11 unverdicted novelty 5.0

    A quantum perceptron with tunable-frequency oscillating activation achieves higher storage capacity than classical perceptrons, but the gain arises solely from the activation function form.