Graphical Models: Representations for Learning, Reasoning and Data Mining
Graphical models are of increasing importance in applied statistics, and in particular in data mining. Providing a self-contained introduction and overview to learning relational, probabilistic, and possibilistic networks from data, this second edition of Graphical Models is thoroughly updated to include the latest research in this burgeoning field, including a new chapter on visualization. The text provides graduate students, and researchers with all the necessary background material,...
Search in google:
The use of graphical models in applied statistics has increased considerably in recent years. At the same time the field of data mining has developed as a response to the large amounts of available data. This book addresses the overlap between these two important areas, highlighting the advantages of using graphical models for data analysis and mining. The Authors focus not only on probabilistic models such as Bayesian and Markov networks but also explore relational and possibilistic graphical models in order to analyse data sets. Presents all necessary background material including uncertainty and imprecision modeling, distribution decomposition and graphical representation.Covers Markov, Bayesian, relational and possibilistic networks.Includes a new chapter on visualization and coverage of clique tree propagation, visualization techniques.Demonstrates learning algorithms based on a large number of different search methods and evaluation measures.Includes a comprehensive bibliography and a detailed index.Features an accompanying website hosting exercises, teaching material and open source software. Researchers and practitioners who use graphical models in their work, graduate students of applied statistics, computer science and engineering will find much of interest in this new edition. Booknews Aimed at researchers as well as graduate students in applied statistics, computer science, and engineering, this text provides an introduction to the use of graphical models for data analysis and data mining. Topics include, for example, conditional independence, learning graphical models from data, inducing a network structure from a database of sample cases, and applications in the telecommunications industry. The authors are with the Otto-von-Guericke-University of Madeburg, Germany. Annotation c. Book News, Inc., Portland, OR (booknews.com)
Preface1Introduction11.1Data and Knowledge11.2Knowledge Discovery and Data Mining51.3Graphical Models101.4Outline of this Book122Imprecision and Uncertainty152.1Modelling Inferences152.2Imprecision and Relational Algebra172.3Uncertainty and Probability Theory192.4Possibility Theory and the Context Model213Decomposition533.1Decomposition and Reasoning543.2Relational Decomposition553.3Probabilistic Decomposition743.4Possibilistic Decomposition823.5Possibility versus Probability874Graphical Representation914.1Conditional Independence Graphs924.2Evidence Propagation in Graphs1155Computing Projections1335.1Databases of Sample Cases1345.2Relational and Sum Projections1355.3Expectation Maximization1375.4Maximum Projections1426Naive Classifiers1516.1Naive Bayes Classifiers1516.2A Naive Possibilistic Classifier1566.3Classifier Simplification1586.4Experimental Results1587Learning Global Structure1617.1Principles of Learning Global Structure1627.2Evaluation Measures1877.3Search Methods2237.4Experimental Results2478Learning Local Structure2538.1Local Network Structure2538.2Learning Local Structure2558.3Experimental Results2599Inductive Causation2619.1Correlation and Causation2619.2Causal and Probabilistic Structure2629.3Stability and Latent Variables2649.4The Inductive Causation Algorithm2669.5Critique of the Underlying Assumptions2679.6Evaluation27210Applications27510.1Application in Telecommunications27510.2Application at Volkswagen27910.3Application at Daimler Chrysler282AProofs of Theorems287B: Software Tools323Bibliography329Index349
\ From the Publisher"All of the necessary background is provided, with material on modeling under uncertainty and imprecision modeling, decomposition of distributions, graphical representation of distributions, applications relating to graphical models, and problems for further research." (Book News, December 2009)\ \ \ \ \ \ Aimed at researchers as well as graduate students in applied statistics, computer science, and engineering, this text provides an introduction to the use of graphical models for data analysis and data mining. Topics include, for example, conditional independence, learning graphical models from data, inducing a network structure from a database of sample cases, and applications in the telecommunications industry. The authors are with the Otto-von-Guericke-University of Madeburg, Germany. Annotation c. Book News, Inc., Portland, OR (booknews.com)\ \
