The effect of priors on Learning with Restricted Boltzmann Machines

Abstract

Restricted Boltzmann Machines (RBMs) are generative models designed to learn from data with a rich underlying structure. In this work, we explore a teacher-student setting where a student RBM learns from examples generated by a teacher RBM, with a focus on the effect of the unit priors on learning efficiency. We consider a parametric class of priors that interpolate between continuous (Gaussian) and binary variables. This approach models various possible choices of visible units, hidden units, and weights for both the teacher and student RBMs. By analyzing the phase diagram of the posterior distribution in both the Bayes optimal and mismatched regimes, we demonstrate the existence of a triple point that defines the critical dataset size necessary for learning through generalization. The critical size is strongly influenced by the properties of the teacher, and thus the data, but is unaffected by the properties of the student RBM. Nevertheless, a prudent choice of student priors can facilitate training by expanding the so-called signal retrieval region, where the machine generalizes effectively.

Accepted for publication in Physica A, Volume 674, 130766. -Added references -Corrected typos -Revised section 3, with Monte Carlo simulations supporting the results (unchanged)

References in corpus (26)

• A high-bias, low-variance introduction to Machine Learning for physicists
• The Nishimori line and Bayesian Statistics
• Constrained Low-rank Matrix Estimation: Phase Transitions, Approximate Message Passing and Applications
• Phase Diagram of Restricted Boltzmann Machines and Generalised Hopfield Networks with Arbitrary Priors
• Efficiency of quantum versus classical annealing in non-convex learning problems
• Phase transitions in Restricted Boltzmann Machines with generic priors
• Extensive load in multitasking associative networks
• Thermodynamics of Restricted Boltzmann Machines and related learning dynamics
• From Dyson to Hopfield: Processing on hierarchical networks
• Spectral Dynamics of Learning Restricted Boltzmann Machines
• How glassy are neural networks?
• Statistical mechanics of unsupervised feature learning in a restricted Boltzmann machine with binary synapses
• Non-Convex Multi-species Hopfield models
• Neural Networks retrieving Boolean patterns in a sea of Gaussian ones
• Hierarchical neural networks perform both serial and parallel processing
• Approximate Survey Propagation for Statistical Inference
• Dense Hebbian neural networks: a replica symmetric picture of supervised learning
• Unsupervised feature learning from finite data by message passing: discontinuous versus continuous phase transition
• Minimal model of permutation symmetry in unsupervised learning
• Inferring effective couplings with Restricted Boltzmann Machines
• Legendre Duality of Spherical and Gaussian Spin Glasses
• Hopfield model with planted patterns: a teacher-student self-supervised learning model
• Dense Hopfield Networks in the Teacher-Student Setting
• Modeling Structured Data Learning with Restricted Boltzmann Machines in the Teacher-Student Setting
• Bayes-optimal limits in structured PCA, and how to reach them
• Fast training and sampling of Restricted Boltzmann Machines

Cited by in corpus (3)

• On the role of non-linear latent features in bipartite generative neural networks
• Saddle Hierarchy in Dense Associative Memory
• Supervised and Unsupervised protocols for hetero-associative neural networks