English

Minimum Conditional Description Length Estimation for Markov Random Fields

Information Theory 2016-02-25 v2 Machine Learning math.IT Statistics Theory Statistics Theory

Abstract

In this paper we discuss a method, which we call Minimum Conditional Description Length (MCDL), for estimating the parameters of a subset of sites within a Markov random field. We assume that the edges are known for the entire graph G=(V,E)G=(V,E). Then, for a subset UVU\subset V, we estimate the parameters for nodes and edges in UU as well as for edges incident to a node in UU, by finding the exponential parameter for that subset that yields the best compression conditioned on the values on the boundary U\partial U. Our estimate is derived from a temporally stationary sequence of observations on the set UU. We discuss how this method can also be applied to estimate a spatially invariant parameter from a single configuration, and in so doing, derive the Maximum Pseudo-Likelihood (MPL) estimate.

Keywords

Cite

@article{arxiv.1602.03061,
  title  = {Minimum Conditional Description Length Estimation for Markov Random Fields},
  author = {Matthew G. Reyes and David L. Neuhoff},
  journal= {arXiv preprint arXiv:1602.03061},
  year   = {2016}
}

Comments

Information Theory and Applications (ITA) workshop, February 2016

R2 v1 2026-06-22T12:46:49.492Z