Inferring a Bayesian Network for Content-Based Image Classification

摘要: Bayesian networks are popular in the classification literature. The simplest kind of network, i.e. naive has gained interest many researchers because quick learning and inferring. However, when there lots classes to be inferred from a similar set evidences, one may prefer have united network. In this paper we present new method for merging order achieve complete network study effect merging. proposed reduces burden A simple measure is also introduced assess stability results after combination classifiers. applied image problem. indicate that addition reduced computation total precision increased alteration each individual class estimable using measure.