Equilibriumandnon-Equilibriumregimesinthe learningofRestrictedBoltzmannMachines
–Neural Information Processing Systems
Inparticular,weshowthat using the popular k (persistent) contrastive divergence approaches, with k small, the dynamics of the learned model are extremely slow and often dominated by strong out-of-equilibrium effects.
Neural Information Processing Systems
Feb-8-2026, 00:23:36 GMT