2 papers
cs.LG2026
Spherical Boltzmann machines: a solvable theory of learning and generation in energy-based models
Thomas Tulinski, Simona Cocco, Rémi Monasson +1
Energy-based models (EBMs) are flexible generative architectures inspired by statistical physics, but their learning and generative properties remain poorly understood. Here, we an…
cond-mat.dis-nn2026
Replica Theory of Spherical Boltzmann Machine Ensembles
Thomas Tulinski, Jorge Fernandez-De-Cossio-Diaz, Simona Cocco +1
Training in machine learning generally consists in finding one model, whose parameters minimize a data-dependent loss. Yet, empirical work shows that ensemble learning, an approach…