Bethe and Related Pairwise Entropy Approximations

Abstract

For undirected graphical models, belief propagation often performs remarkably well for approximate marginal inference, and may be viewed as a heuristic to minimize the Bethe free energy. Focusing on binary pairwise models, we demonstrate that several recent results on the Bethe approximation may be generalized to a broad family of related pairwise free energy approximations with arbitrary counting numbers. We explore comparisons to the true (Gibbs) free energy and shed light on the empirical success of the Bethe approximation.

Cite

Text

Weller. "Bethe and Related Pairwise Entropy Approximations." Conference on Uncertainty in Artificial Intelligence, 2015.

Markdown

[Weller. "Bethe and Related Pairwise Entropy Approximations." Conference on Uncertainty in Artificial Intelligence, 2015.](https://mlanthology.org/uai/2015/weller2015uai-bethe/)

BibTeX

@inproceedings{weller2015uai-bethe,
  title     = {{Bethe and Related Pairwise Entropy Approximations}},
  author    = {Weller, Adrian},
  booktitle = {Conference on Uncertainty in Artificial Intelligence},
  year      = {2015},
  pages     = {942-951},
  url       = {https://mlanthology.org/uai/2015/weller2015uai-bethe/}
}