Sensitivity Analysis of Probabilistic Graphical Models

Probabilistic graphical models such as Bayesian
networks are widely used for large-scale data
analysis in various fields such as customer data
analysis and medical diagnosis, as they model
probabilistic knowledge naturally and allow the use
of efficient inference algorithms to draw
conclusions from the model. Sensitivity analysis of
probabilistic graphical models is the analysis of
the relationships between the inputs (local
beliefs), such as network parameters, and the
outputs (global beliefs), such as values of
probabilistic queries, and addresses the central
research problem of how beliefs will be changed when
we incorporate new information to the current model.
This book provides many theoretical results, such as
the assessment of global belief changes due to local
belief changes, the identification of local belief
changes that induce certain global belief changes,
and the quantifying of belief changes in general.
These results can be applied on the modeling and
inference of Bayesian networks, and provide a
critical tool for the researchers, developers,
and users of Bayesian networks during the process of
probabilistic data modeling and reasoning.

Autorentext

Hei Chan is a research fellow at the Center for Service Research of the National Institute of Advanced Industrial Science and Technology (AIST), in Tokyo, Japan. He received his Ph.D. degree in computer science at the University of California, Los Angeles (UCLA). His research interests include data modeling, Bayesian networks, and causality.

Klappentext

Probabilistic graphical models such as Bayesian networks are widely used for large-scale data analysis in various fields such as customer data analysis and medical diagnosis, as they model probabilistic knowledge naturally and allow the use of efficient inference algorithms to draw conclusions from the model. Sensitivity analysis of probabilistic graphical models is the analysis of the relationships between the inputs (local beliefs), such as network parameters, and the outputs (global beliefs), such as values of probabilistic queries, and addresses the central research problem of how beliefs will be changed when we incorporate new information to the current model. This book provides many theoretical results, such as the assessment of global belief changes due to local belief changes, the identification of local belief changes that induce certain global belief changes, and the quantifying of belief changes in general. These results can be applied on the modeling and inference of Bayesian networks, and provide a critical tool for the researchers, developers, and users of Bayesian networks during the process of probabilistic data modeling and reasoning.