AI RESEARCH
Spherical Boltzmann machines: a solvable theory of learning and generation in energy-based models
arXiv CS.LG
•
ArXi:2605.09031v1 Announce Type: new Energy-based models (EBMs) are flexible generative architectures inspired by statistical physics, but their learning and generative properties remain poorly understood. Here, we analyze a solvable EBM in the high-dimensional limit: the spherical Boltzmann machine (SBM). Combining tools from random matrix theory and dynamical mean-field theory, we: solve exact equations describing the