Издательство InTech, 2010, -442 pp.
Bayesian networks are graphical models that represent the probabilistic relationships among a large number of variables and perform probabilistic inference with those variables. They constitute a formal framework for the representation and communication of decisions resulting from reasoning under uncertainty. Bayesian networks, which were named after Thomas Bayes (1702-1761), one of the founders of the probability theory, have emerged from several mathematical researches made in the 1980s, and particularly from works on belief networks, causal networks and influence diagrams.
Bayesian networks were first known in the 1990s as Probabilistic Expert Systems, inspired by the seminal book of Judea Pearl (1988), who was a pioneer of the probabilistic approach to artificial intelligence and is referred to as the founder of Bayesian networks. Bayesian networks are thus at least 22 years old and during the last two decades a lot of work has beendone on leaing and inference with Bayesian networks. The last ten years particularly saw a massive increase in the application of BN to real-world problems, including diagnosis, forecasting, manufacturing control, information retrieval, prediction and even planning. Almost all scientific and technical fields have seen the successful use of BN as a tool for modelling the complex relationships among a large number of variables and for doing inference. The most recent applications have been in information and communications technologies, biomedicine, genomics and bioinformatics.
The first decade of this new millennium saw the emergence of excellent algorithms for leaing Bayesian networks from data and for doing inference in Bayesian networks and influence diagrams. According to Google Scholar, the number of research papers and technical reports on Bayesian networks is over fifty thousand and at least seven specific books on Bayesian networks were published in 2009.
Despite this abundance of literature, there is still a need for specialized books that present original contributions both in methodology and applications of Bayesian networks. This book emphasizes these two aspects and is intended for users (current or potential) of Bayesian networks in both academic institutions (researchers, teachers, students) and industry (engineers, analysts, etc.) who want to stay up to date with Bayesian network algorithms and technologies and their use in building probabilistic expert systems and modelling complex systems.
The book is organized in two major parts. The first part, extending from chapter 1 to 10, mainly deals with theory and algorithms for leaing and inference in Bayesian networks. The second part, composed of all subsequent chapters, gives selected applications of Bayesian networks in several fields, including fault diagnosis, information technology, telecommunication networks, traffic flow, building design and biology.
The book chapters are original manuscripts written by experienced researchers that have made significant contributions to the field of Bayesian networks. Although all chapters are self- contained, the reader should be familiar with texts written in mathematical and statistical language to gain full benefit from the book. I am convinced that this book will be a very useful tool for everyone who is conceed with modelling systems containing causality with inherent uncertainty and I hope that readers will find not only technical aspects for using and implementing Bayesian networks to solve their problem, but also new ideas on how their current research and work can benefit from one of the major tools of the 21st century.
Bayesian networks are graphical models that represent the probabilistic relationships among a large number of variables and perform probabilistic inference with those variables. They constitute a formal framework for the representation and communication of decisions resulting from reasoning under uncertainty. Bayesian networks, which were named after Thomas Bayes (1702-1761), one of the founders of the probability theory, have emerged from several mathematical researches made in the 1980s, and particularly from works on belief networks, causal networks and influence diagrams.
Bayesian networks were first known in the 1990s as Probabilistic Expert Systems, inspired by the seminal book of Judea Pearl (1988), who was a pioneer of the probabilistic approach to artificial intelligence and is referred to as the founder of Bayesian networks. Bayesian networks are thus at least 22 years old and during the last two decades a lot of work has beendone on leaing and inference with Bayesian networks. The last ten years particularly saw a massive increase in the application of BN to real-world problems, including diagnosis, forecasting, manufacturing control, information retrieval, prediction and even planning. Almost all scientific and technical fields have seen the successful use of BN as a tool for modelling the complex relationships among a large number of variables and for doing inference. The most recent applications have been in information and communications technologies, biomedicine, genomics and bioinformatics.
The first decade of this new millennium saw the emergence of excellent algorithms for leaing Bayesian networks from data and for doing inference in Bayesian networks and influence diagrams. According to Google Scholar, the number of research papers and technical reports on Bayesian networks is over fifty thousand and at least seven specific books on Bayesian networks were published in 2009.
Despite this abundance of literature, there is still a need for specialized books that present original contributions both in methodology and applications of Bayesian networks. This book emphasizes these two aspects and is intended for users (current or potential) of Bayesian networks in both academic institutions (researchers, teachers, students) and industry (engineers, analysts, etc.) who want to stay up to date with Bayesian network algorithms and technologies and their use in building probabilistic expert systems and modelling complex systems.
The book is organized in two major parts. The first part, extending from chapter 1 to 10, mainly deals with theory and algorithms for leaing and inference in Bayesian networks. The second part, composed of all subsequent chapters, gives selected applications of Bayesian networks in several fields, including fault diagnosis, information technology, telecommunication networks, traffic flow, building design and biology.
The book chapters are original manuscripts written by experienced researchers that have made significant contributions to the field of Bayesian networks. Although all chapters are self- contained, the reader should be familiar with texts written in mathematical and statistical language to gain full benefit from the book. I am convinced that this book will be a very useful tool for everyone who is conceed with modelling systems containing causality with inherent uncertainty and I hope that readers will find not only technical aspects for using and implementing Bayesian networks to solve their problem, but also new ideas on how their current research and work can benefit from one of the major tools of the 21st century.