English

Entropy Message Passing

Machine Learning 2016-11-18 v6 Information Theory math.IT

Abstract

The paper proposes a new message passing algorithm for cycle-free factor graphs. The proposed "entropy message passing" (EMP) algorithm may be viewed as sum-product message passing over the entropy semiring, which has previously appeared in automata theory. The primary use of EMP is to compute the entropy of a model. However, EMP can also be used to compute expressions that appear in expectation maximization and in gradient descent algorithms.

Keywords

Cite

@article{arxiv.0906.2895,
  title  = {Entropy Message Passing},
  author = {Velimir M. Ilic and Miomir S. Stankovic and Branimir T. Todorovic},
  journal= {arXiv preprint arXiv:0906.2895},
  year   = {2016}
}

Comments

5 pages, 1 figure, to appear in IEEE Transactions on Information Theory