FEM is a conditional energy model for hybrid Bayesian networks that uses learned embeddings and valley regularization to enable accurate multimodal posterior inference and compositional sampling.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Free Energy Manifold: Score-Based Inference for Hybrid Bayesian Networks
FEM is a conditional energy model for hybrid Bayesian networks that uses learned embeddings and valley regularization to enable accurate multimodal posterior inference and compositional sampling.