Variable elimination (VE) is a simple and general exact inference algorithm in probabilistic graphical models, such as Bayesian networks and Markov random fields. It can be used for inference of maximum a posteriori (MAP) state or estimation of marginal distribution over a subset of variables. The algorithm has exponential time complexity, but could be efficient in practice for the low-treewidth graphs, if the proper elimination order is used.
Enabling a key reduction in algorithmic complexity, a factor , also known as a potential, of variables is a relation between each instantiation of of variables to a non-negative number, commonly denoted as . A factor does not necessarily have a set interpretation. One may perform operations on factors of different representations such as a probability distribution or conditional distribution. Joint distributions often become too large to handle as the complexity of this operation is exponential. Thus variable elimination becomes more feasible when computing factorized entities.
Algorithm 1, called sum-out (SO), or marginalization, eliminates a single variable from a set of factors, and returns the resulting set of factors. The algorithm collect-relevant simply returns those factors in involving variable .
Algorithm 1 sum-out(,)
- = collect factors relevant to
- = the product of all factors in
Here we have a joint probability distribution. A variable, can be summed out between a set of instantiations where the set at minimum must agree over the remaining variables. The value of is irrelevant when it is the variable to be summed out. 
After eliminating , its reference is excluded and we are left with a distribution only over the remaining variables and the sum of each instantiation.
The resulting distribution which follows the sum-out operation only helps to answer queries that do not mention . Also worthy to note, the summing-out operation is commutative.
Computing a product between multiple factors results in a factor compatible with a single instantiation in each factor.
Algorithm 2 mult-factors(,)
- = Union of all variables between product of factors
- = a factor over where for all
- For each instantiation
- For 1 to
- instantiation of variables consistent with
- For 1 to
Factor multiplication is not only commutative but also associative.
The most common query type is in the form where and are disjoint subsets of , and is observed taking value . A basic algorithm to computing p(X|E = e) is called variable elimination (VE), first put forth in.
Taken from, this algorithm computes from a discrete Bayesian network B. VE calls SO to eliminate variables one by one. More specifically, in Algorithm 2, is the set C of CPTs for B, is a list of query variables, is a list of observed variables, is the corresponding list of observed values, and is an elimination ordering for variables , where denotes .
Variable Elimination Algorithm VE()
- Multiply factors with appropriate CPTs While σ is not empty
- Remove the first variable from
- = sum-out
- = the product of all factors
Finding the optimal order in which to eliminate variables is an NP-hard problem. As such there are heuristics one may follow to better optimize performance by order:
- Minimum Degree: Eliminate the variable which results in constructing the smallest factor possible.
- Minimum Fill: By constructing an undirected graph showing variable relations expressed by all CPTs, eliminate the variable which would result in the least edges to be added post elimination.
- Zhang, N.L., Poole, D.:A Simple Approach to Bayesian Network Computations.In: 7th Canadian Conference on Artificial Intelligence,pp. 171--178. Springer, New York (1994)
- Darwiche, Adnan (2009-01-01). Modeling and Reasoning with Bayesian Networks. doi:10.1017/cbo9780511811357. ISBN 9780511811357.
- Koller, D., Friedman, N.: Probabilistic Graphical Models: Principles and Techniques. MIT Press, Cambridge, MA (2009)
|This statistics-related article is a stub. You can help Wikipedia by expanding it.|
|This computer science article is a stub. You can help Wikipedia by expanding it.|