• Corpus ID: 811964

# Hierarchical Models, Marginal Polytopes, and Linear Codes

```@article{Kahle2009HierarchicalMM,
title={Hierarchical Models, Marginal Polytopes, and Linear Codes},
author={Thomas Kahle and Walter Wenzel and Nihat Ay},
journal={Kybernetika},
year={2009},
volume={45},
pages={189-207}
}```
• Published 1 April 2008
• Computer Science
• Kybernetika
In this paper, we explore a connection between binary hierarchical models, their marginal polytopes and codeword polytopes, the convex hulls of linear codes. The class of linear codes that are realizable by hierarchical models is determined. We classify all full dimensional polytopes with the property that their vertices form a linear code and give an algorithm that determines them.
11 Citations

## Figures and Tables from this paper

Permutation Polytopes of Cyclic Groups
• Mathematics
• 2011
We investigate the combinatorics and geometry of permutation polytopes associated to cyclic permutation groups, i.e., the convex hulls of cyclic groups of permutation matrices. We give formulas for
Neighborliness of Marginal Polytopes
A neighborliness property of marginal polytopes of hierar- chical models, depending on the cardinality of the smallest non-face of the underlying simplicial complex, is shown. The case of binary
Mixture decompositions of exponential families - using a decomposition of their sample spaces
An approach based on coverings and packings of the face lattice of the corresponding convex support polytopes and results from coding theory is proposed and shown that \$m=q^{N-1}\$ is the smallest number for which any distribution of \$N\$-ary variables can be written as mixture of \$m\$ independent \$q\$-ARY variables.
Dimension of Marginals of Kronecker Product Models; Geometry of hidden-visible products of exponential families
• Mathematics
• 2015
A Kronecker product model is the set of visible marginal probability distributions of an exponential family whose sufficient statistics matrix factorizes as a Kronecker product of two matrices, one
Dimension of Marginals of Kronecker Product Models
• Mathematics
SIAM J. Appl. Algebra Geom.
• 2017
It is proved that the binary restricted Boltzmann machine always has the expected dimension, equal to the minimum of the number of natural parameters and the dimension of the ambient probability simplex.
When Does a Mixture of Products Contain a Product of Mixtures?
• Computer Science
SIAM J. Discret. Math.
• 2015
It is shown that a mixture of products requiring an exponentially larger number of parameters is needed to represent the probability distributions which can be obtained as products of mixtures.
Spectral Hypergraph Partitioning and Relative Entropy
This thesis is about the information-theoretic difference between hypergraphs and undirected graphs, which arises in spectral bisections of these objects. We also investigate the effect of general
Maximizing the Divergence from a Hierarchical Model of Quantum States
• Computer Science
Open Syst. Inf. Dyn.
• 2015
This work studies many-party correlations quantified in terms of the Umegaki relative entropy (divergence) from a Gibbs family known as a hierarchical model and discusses global maximizers of the mutual information of separable qubit states.

## References

SHOWING 1-10 OF 35 REFERENCES
Gröbner Bases and Polyhedral Geometry of Reducible and Cyclic Models
• Mathematics, Computer Science
J. Comb. Theory, Ser. A
• 2002
The polyhedral structure and combinatorics of polytopes that arise from hierarchical models in statistics are studied, and it is shown how to construct Grobner bases of toric ideals associated to a subset of such models.
Neighborliness of Marginal Polytopes
A neighborliness property of marginal polytopes of hierar- chical models, depending on the cardinality of the smallest non-face of the underlying simplicial complex, is shown. The case of binary
On the toric algebra of graphical models
• Mathematics
• 2006
We formulate necessary and sufficient conditions for an arbitrary discrete probability distribution to factor according to an undirected graphical model, or a log-linear model, or other more general
polymake: a Framework for Analyzing Convex Polytopes
• Computer Science
• 2000
An overview of the functionality as well as of the structure of the polymake tool is given, seen as a first approximation to a polymake handbook.
Using Linear Programming to Decode Linear Codes
• Computer Science
• 2003
A method for approximate ML decoding of an arbi- trary binary linear code, based on a linear program- ming (LP) relaxation that is dened by a factor graph or parity check representation of the code, and it is proved that the performance of LP decoding is equivalent to standard iterative decoding.
Lectures on 0/1-Polytopes
Several aspects of the complexity of higher-dimensional 0/1-polytopes are studied: the doubly-exponential number of combinatorial types, the number of facets which can be huge, and the coefficients of defining inequalities which sometimes turn out to be extremely large.
Support sets of distributions with given interaction structure
• Mathematics
• 2006
We study closures of hierarchical models which are exponential families associated with hypergraphs by decomposing the corresponding interaction spaces in a natural and transparent way. Here, we
Maximizing multi-information
• Mathematics
Kybernetika
• 2006
Low-dimensional exponential families that contain the maximizers of stochastic interdependence in their closure are investigated and it is obtained that the exponential family of pure pair-interactions contains all global maximizer of the multi-information in its closure.
Variational inference in graphical models: The view from the marginal polytope
• Computer Science
• 2008
Taking the “zero-temperature limit” recovers a variational representation for MAP computation as a linear program (LP) over the marginal polytope, which clarifies the essential ingredients of known variational methods, and also suggests novel relaxations.
Information geometry on hierarchy of probability distributions
• S. Amari
• Computer Science, Mathematics
IEEE Trans. Inf. Theory
• 2001
The orthogonal decomposition of an exponential family or mixture family of probability distributions has a natural hierarchical structure is given and is important for extracting intrinsic interactions in firing patterns of an ensemble of neurons and for estimating its functional connections.