Model-based hierarchical clustering algorithm using Bayesian networks

Student thesis: Master thesis (including HD thesis)

  • Rasa Jurgelenaite
This report introduces and empirically evaluates a model-based clustering algorithm, which uses directed graphical models (Bayesian networks) to describe the clusters and adopts hierarchical clustering strategy. Two model structure learning algorithms, the PC algorithm and the KES algorithm, are applied in the proposed clustering algorithm. The experimental results for gene expression data are reported evidencing the reasonable perfomance of the introduced approach.
Publication dateJun 2003
ID: 61058212