|
Abstract:
In this paper we address the problem of learning the
structure in nonlinear Markov networks with continuous variables.
Markov networks are well suited to model relationships which do not
exhibit a natural causal ordering. We represent the quantitative
relationships between variables using neural networks as models for
conditional probability densities. This approach is well suited for
inference by Gibbs sampling. Using a financial and a sociological
data set we show that interesting structures can be found using our
approach.
|
