Comparing Hierarchical Markov Networks and Multiply Sectioned Bayesian Networks

Abstract. Multiply sectioned Bayesian networks (MSBNs) were originally proposed as a modular representation of uncertain knowledge by sectioning a large Bayesian network (BN) into smaller units. More recently, hierarchical Markov networks (HMNs) were developed in part as an hierarchical representation of the flat BN. In this paper, we compare the MSBN and HMN representations. The MSBN representation does not specify how to section a BN, nor is it a faithful representation of BNs. On the contrary, a given BN has a unique HMN representation, which encodes precisely those independencies encoded in the BN. More importantly, we show that failure to encode known independencies can lead to unnecessary computation in the MSBN representation. These results, in particular, suggest that HMNs may be a more natural representation of BNs than MSBNs.

• Type: Conference
• Year: 2003
• Where: ISMIS
• Authors: Cory J. Butz, H. Geng
• Keywords: Bayesian Network, Hmn Representation, ISMIS 2003, MSBN Representation