Learning Bayesian Network Structure from英文文献.pdfVIP

Learning Bayesian Network Structure from Incomplete Data without Any Assumption 1 2 C´eline Fiot , G.A. Putri Saptawati , Anne Laurent1, and Maguelonne Teisseire1 1 LIRMM - Univ. Montpellier II, CNRS 161 rue Ada, 34392 Montpellier, France {fiot,laurent,teisseire}@lirmm.fr 2 Institut Teknologi Bandung Jl. Ganesha 10, Bandung 40132, Indonesia putri@informatika.org Abstract. Since most real-life data contain missing values, reasoning and learning with incomplete data has become crucial in data mining and machine learning. In particular, Bayesian networks are one machine learning technique that allows for reasoning with incomplete data, but training such networks on incomplete data may be a difficult task. Many methods were thus proposed to learn Bayesian network structure from incomplete data, based on multiple structure generation and scoring of their adequacy to the dataset. However, this kind of approaches may be time-consuming. Therefore we propose an efficient dependency analysis approach that uses a redefinition of probability calculation to take incom- plete records into account while learning BN structure, without generat- ing multiple possibilities. Some experiments on well-known benchmarks are described to show the validity of our proposal. 1 Introduction Graphical models [1] are tools combining two different areas: graph theory and probability theory. They are often used to illustrate and work with conditional independencies and probabilistic relationships between variables i