Abstract:Restricted Boltzmann machines (RBMs) constitute one of the main models for machine statistical inference and they are widely employed in Artificial Intelligence as powerful tools for (deep) learning. However, in contrast with countless remarkable practical successes, their mathematical formalization has been largely elusive: from a statistical-mechanics perspective these systems display the same (random) Gibbs measure of bi-partite spin-glasses, whose rigorous treatment is notoriously difficult. In this work, beyond providing a brief review on RBMs from both the learning and the retrieval perspectives, we aim to contribute to their analytical investigation, by considering two distinct realizations of their weights (i.e., Boolean and Gaussian) and studying the properties of their related free energies. More precisely, focusing on a RBM characterized by digital couplings, we first extend the Pastur-Shcherbina-Tirozzi method (originally developed for the Hopfield model) to prove the self-averaging property for the free energy, over its quenched expectation, in the infinite volume limit, then we explicitly calculate its simplest approximation, namely its annealed bound. Next, focusing on a RBM characterized by analogical weights, we extend Guerra's interpolating scheme to obtain a control of the quenched free-energy under the assumption of replica symmetry: we get self-consistencies for the order parameters (in full agreement with the existing Literature) as well as the critical line for ergodicity breaking that turns out to be the same obtained in AGS theory. As we discuss, this analogy stems from the slow-noise universality. Finally, glancing beyond replica symmetry, we analyze the fluctuations of the overlaps for an estimate of the (slow) noise affecting the retrieval of the signal, and by a stability analysis we recover the Aizenman-Contucci identities typical of glassy systems.

Replica analysis of Bayesian data clustering

Replica analysis of the lattice-gas Restricted Boltzmann Machine partition function

A Statistical Clustering Model and Algorithm

Replica Analysis of the Linear Model with Markov or Hidden Markov Signal Priors

Strong replica symmetry in high-dimensional optimal Bayesian inference

Energy-Based Clustering: Fast and Robust Clustering of Data with Known Likelihood Functions

Bayesian approach to clustering real value, categorical and network data: solution via variational methods

Clustering of heterogeneous populations of networks

Replica Symmetry Breaking in Random Non-Hermitian Systems

Free energies of Boltzmann Machines: self-averaging, annealed and replica symmetric approximations in the thermodynamic limit

The metastate approach to thermodynamic chaos

A model-based approach for clustering binned data

Bayesian ensemble refinement by replica simulations and reweighting

Replica symmetry breaking in supervised and unsupervised Hebbian networks

Replica trick with real replicas: A way to build in thermodynamic homogeneity

Significance analysis and statistical mechanics: an application to clustering

The adaptive interpolation method: a simple scheme to prove replica formulas in Bayesian inference

A review on Bayesian model-based clustering

Frequency decomposition of astrometric signature of planetary systems

Replica Analysis for Generalized Linear Regression with IID Row Prior

Rademacher complexity and spin glasses: A link between the replica and statistical theories of learning