Probabilistic Reasoning in Multiagent Systems: A Graphical Models Approach

by Yang Xiang

• ISBN13: 9780521813082
• ISBN10: 0521813085
• Format: Hardcover
• Copyright: 2002-08-26
• Publisher: Cambridge University Press

Summary

This book investigates the opportunities in building intelligent decision support systems offered by multi-agent distributed probabilistic reasoning. Probabilistic reasoning with graphical models, also known as Bayesian networks or belief networks, has become an active field of research and practice in artificial intelligence, operations research and statistics in the last two decades. The success of this technique in modeling intelligent decision support systems under the centralized and single-agent paradigm has been striking. In this book, the author extends graphical dependence models to the distributed and multi-agent paradigm. He identifies the major technical challenges involved in such an endeavor and presents the results from a decade's research. The framework developed in the book allows distributed representation of uncertain knowledge on a large and complex environment embedded in multiple cooperative agents, and effective, exact and distributed probabilistic inference.

Author Biography

Yang Xiang is Associate Professor of Computing and Information Science at the University of Guelph, Canada, where he directs the Intelligent Decision Support System Laboratory

Table of Contents

Preface	ix
Introduction	1	(15)
Intelligent Agents	1	(3)
Reasoning about the Environment	4	(1)
Why Uncertain Reasoning?	5	(2)
Multiagent Systems	7	(4)
Cooperative Multiagent Probabilistic Reasoning	11	(2)
Application Domains	13	(1)
Bibliographical Notes	14	(2)
Bayesian Networks	16	(21)
Guide to Chapter 2	16	(3)
Basics on Bayesian Probability Theory	19	(4)
Belief Updating Using JPD	23	(1)
Graphs	24	(3)
Bayesian Networks	27	(3)
Local Computation and Message Passing	30	(1)
Message Passing over Multiple Networks	31	(2)
Approximation with Massive Message Passing	33	(2)
Bibliographical Notes	35	(1)
Exercises	36	(1)
Belief Updating and Cluster Graphs	37	(24)
Guide to Chapter 3	38	(2)
Cluster Graphs	40	(3)
Conventions for Message Passing in Cluster Graphs	43	(1)
Relation with λ - π Message Passing	44	(3)
Message Passing in Nondegenerate Cycles	47	(6)
Message Passing in Degenerate Cycles	53	(3)
Junction Trees	56	(3)
Bibliographical Notes	59	(1)
Exercises	59	(2)
Junction Tree Representation	61	(25)
Guide to Chapter 4	62	(2)
Graphical Separation	64	(4)
Sufficient Message and Independence	68	(1)
Encoding Independence in Graphs	69	(2)
Junction Trees and Chordal Graphs	71	(5)
Triangulation by Elimination	76	(2)
Junction Trees as I-maps	78	(2)
Junction Tree Construction	80	(3)
Bibliographical Notes	83	(1)
Exercises	84	(2)
Belief Updating with Junction Trees	86	(21)
Guide to Chapter 5	86	(2)
Algebraic Properties of Potentials	88	(6)
Potential Assignment in Junction Trees	94	(3)
Passing Belief over Separators	97	(3)
Passing Belief through a Junction Tree	100	(4)
Processing Observations	104	(1)
Bibliographical Notes	105	(1)
Exercises	105	(2)
Multiply Sectioned Bayesian Networks	107	(35)
Guide to Chapter 6	108	(4)
The Task of Distributed Uncertain Reasoning	112	(5)
Organization of Agents during Communication	117	(7)
Agent Interface	124	(4)
Multiagent Dependence Structure	128	(5)
Multiply Sectioned Bayesian Networks	133	(4)
Bibliographical Notes	137	(3)
Exercises	140	(2)
Linked Junction Forests	142	(40)
Guide to Chapter 7	143	(3)
Multiagent Distributed Compilation of MSBNs	146	(1)
Multiagent Moralization of MSDAG	147	(5)
Effective Communication Using Linkage Trees	152	(3)
Linkage Trees as I-maps	155	(3)
Multiagent Triangulation	158	(16)
Constructing Local Junction Trees and Linkage Trees	174	(7)
Bibliographical Notes	181	(1)
Exercises	181	(1)
Distributed Multiagent Inference	182	(33)
Guide to Chapter 8	183	(3)
Potentials in a Linked Junction Forest	186	(4)
Linkage Trees over the Same d-sepset	190	(2)
Extended Linkage Potential	192	(2)
E-message Passing between Agents	194	(2)
Multiagent Communication	196	(5)
Troubleshooting a Digital System	201	(6)
Complexity of Multiagent Communication	207	(1)
Regional Multiagent Communication	208	(1)
Alternative Methods for Multiagent Inference	209	(3)
Bibliographical Notes	212	(1)
Exercises	213	(2)
Model Construction and Verification	215	(59)
Guide to Chapter 9	216	(1)
Multiagent MSBN System Integration	217	(2)
Verification of Subdomain Division	219	(2)
Agent Registration	221	(2)
Distributed Verification of Acyclicity	223	(14)
Verification of Agent Interface	237	(34)
Complexity of Cooperative d-sepset Testing	271	(1)
Bibliographical Notes	272	(1)
Exercises	272	(2)
Looking into the Future	274	(13)
Multiagent Reasoning in Dynamic Domains	274	(3)
Multiagent Decision Making	277	(2)
What If Verification Fails?	279	(1)
Dynamic Formation of MSBNs	279	(1)
Knowledge Adaptation and Learning	280	(1)
Negotation over Agent Interfaces	281	(2)
Relaxing Hypertree Organization	283	(1)
Model Approximation	284	(1)
Mixed Models	285	(1)
Bibliographical Notes	285	(2)
Bibliography	287	(6)
Index	293