English

Deep Boltzmann machines: rigorous results at arbitrary depth

Mathematical Physics 2021-08-04 v1 Disordered Systems and Neural Networks math.MP

Abstract

A class of deep Boltzmann machines is considered in the simplified framework of a quenched system with Gaussian noise and independent entries. The quenched pressure of a KK-layers spin glass model is studied allowing interactions only among consecutive layers. A lower bound for the pressure is found in terms of a convex combination of KK Sherrington-Kirkpatrick models and used to study the annealed and replica symmetric regimes of the system. A map with a one dimensional monomer-dimer system is identified and used to rigorously control the annealed region at arbitrary depth KK with the methods introduced by Heilmann and Lieb. The compression of this high noise region displays a remarkable phenomenon of localisation of the processing layers. Furthermore a replica symmetric lower bound for the limiting quenched pressure of the model is obtained in a suitable region of the parameters and the replica symmetric pressure is proved to have a unique stationary point.

Keywords

Cite

@article{arxiv.2004.04495,
  title  = {Deep Boltzmann machines: rigorous results at arbitrary depth},
  author = {Diego Alberici and Pierluigi Contucci and Emanuele Mingione},
  journal= {arXiv preprint arXiv:2004.04495},
  year   = {2021}
}

Comments

28 pages

R2 v1 2026-06-23T14:45:27.786Z