Using Cooperative Coevolution for Data Mining of Bayesian Networks - Advanced Techniques in Knowledge Discovery and Data Mining - page 153

6. Using CooperativeCoevolution for Data
Mining of Bayesian Networks
Man Leung Wong, 1 Shing Yan Lee, 2 and Kwong Sak Leung 2
1 Department of Information Systems, Lingnan University, Tuen Mun, Hong
Kong.
2 Department of Computer Science and Engineering, The Chinese University
of Hong Kong, Shatin, Hong Kong.
Bayesian networks are formal knowledge representation tools that provide
reasoning under uncertainty. The applications of Bayesian networks are
widespread, including data mining, information retrieval, and various diag-
nostic systems. Although Bayesian networks are useful, the learning problem,
namely to construct a network automatically from data, remains a di cult
problem. Recently, some researchers have adopted evolutionary computation
for learning. However, the drawback is that the approach is slow. In this
chapter, we propose a hybrid framework for Bayesian network learning. By
combining the merits of two different learning approaches, we expect an im-
provement in learning speed. In brief, the new learning algorithm consists
of two phases: the conditional independence (CI) test phase and the search
phase. In the CI test phase, we conduct dependency analysis, which helps
to reduce the search space. In the search phase, we perform model search-
ing using an evolutionary approach, called cooperative coevolution. When
comparing our new algorithm with an existing algorithm, we find that our
algorithm performs faster and is more accurate in many cases.
6.1 Introduction
Bayesian networks, or Bayesian belief networks, are popular in dealing with
uncertainty for designing intelligent systems. Basically, a Bayesian network
is a graph that depicts conditional independence among random variables
in the domain. In Fig. 6.1, a Bayesian network example is shown. By defi-
nition, a Bayesian network also encodes the joint probability distribution of
the random variables. With a network at hand, we can perform probabilistic
inference for various uses. For instance, we can predict the most likely out-
come of certain variables based on the observation of others. In light of this,
Bayesian networks are widely used in diagnostic systems. For example, in
medical diagnosis, there is MUNIN, which is used for diagnosing diseases in
muscles and nerves, and PATHFINDER, which is used for diagnosing lymph
node disease [6.1]. They are also used in information retrieval [6.2] and printer
troubleshooting [6.3].
The literature on Bayesian networks concentrates on two major issues: the
learning problem and the inference problem. Here, we focus our attention on