Next: Marginalization, inference and sampling
Up: Learning with Mixtures of
Previous: Related work
Tree distributions
In this section we introduce the tree model and the notation that
will be used throughout the paper. Let denote a set of
discrete random variables of interest. For each random variable
let represent its range,
a
particular value, and the (finite) cardinality of .
For each subset of , let
and let denote an assignment to the variables in .
To simplify notation will be denoted by and
will be denoted simply by . Sometimes we need to refer to the
maximum of over ; we denote this value by .
We begin with undirected (Markov random field) representations of tree
distributions. Identifying the vertex set of a graph with the set of
random variables , consider a graph , where is a
set of undirected edges. We allow a tree to have multiple
connected components (thus our ``trees'' are generally called
forests). Given this definition, the number of edges and
the number of connected components are related as follows:
implying that adding an edge to a tree reduces the number of
connected components by 1. Thus, a tree can have at most
edges. In this latter case we refer to the tree as a spanning
tree.
We parameterize a tree in the following way. For and
, let denote a joint probability distribution on
and . We require these distributions to be consistent with
respect to marginalization, denoting by the marginal of
or
with respect to for any
. We now assign a distribution to the graph
as follows:
|
(1) |
where deg is the degree of vertex ; i.e., the number of
edges incident to . It can be verified that is in
fact a probability distribution; moreover, the pairwise probabilities
are the marginals of .
A tree distribution is defined to be any distribution
that admits a factorization of the form (1).
Tree distributions can also be represented using directed (Bayesian
network) graphical models. Let denote a directed
tree (possibly a forest), where is a set of directed edges and
where each node has (at most) one parent, denoted pa.
We parameterize this graph as follows:
|
(2) |
where
is an arbitrary
conditional distribution. It can be verified that indeed
defines a probability distribution; moreover, the marginal
conditionals of are given by the conditionals
.
We shall call the representations (1) and
(2) the undirected and directed tree
representations of the distribution respectively. We can
readily convert between these representations; for example, to
convert (1) to a directed representation we choose
an arbitrary root in each connected component and direct each
edge away from the root. For with closer to
the root than , let pa. Now compute the conditional
probabilities corresponding to each directed edge by recursively
substituting
by
starting from the root.
Figure 2 illustrates this process on a tree with
5 vertices.
Figure 2:
A tree in its undirected (a) and directed (b) representations.
|
The directed tree representation has the advantage of having
independent parameters. The total number of free parameters in either
representation is:
|
(3) |
The right-hand side of (3) shows that each edge
increases the number of parameters by
.
The set of conditional independencies associated with a tree
distribution are readily characterized [Lauritzen, Dawid, Larsen, Leimer
1990].
In particular, two subsets are independent given
if intersects every path (ignoring the direction
of edges in the directed case) between and for all
and .
Subsections
Next: Marginalization, inference and sampling
Up: Learning with Mixtures of
Previous: Related work
Journal of Machine Learning Research
2000-10-19