Discovering weakly-interacting factors in a complex stochastic process

Dynamic Bayesian networks are structured representations of stochastic pro-cesses. Despite their structure, exact inference in DBNs is generally intractable. One approach to approximate inference involves grouping the variables in the process into smaller factors and keeping independent beliefs over these factors. In this paper we present several techniques for decomposing a dynamic Bayesian network automatically to enable factored inference. We examine a number of fea-tures of a DBN that capture different types of dependencies that will cause error in factored inference. An empirical comparison shows that the most useful of these is a heuristic that estimates the mutual information introduced between factors by one step of belief propagation. In addition to features computed over entire factors, for efficiency we explored scores computed over pairs of variables. We present search methods that use these features, pairwise and not, to find a factor-ization, and we compare their results on several datasets. Automatic factorization extends the applicability of factored inference to large, complex models that are undesirable to factor by hand. Moreover, tests on real DBNs show that automatic factorization can achieve significantly lower error in some cases.