17 February 2010
Venue: Bancroft Road Teaching Room 3.02, Mile End
'JChain Event Graphs: A new graphical framework for asymmetric modeling'
Speaker: Professor Jim Smith, University of Warwick
Bayesian Networks have proved to be a very useful tool for expressing and making inferences about high dimensional problems. However experience has taught us that not all problems, especially highly asymmetric ones, are easily or naturally expressible within this framework. The chain event graph is a graph which is much more expressive than the finite discrete Bayesian Network. Like the Bayesian Network it can also be used as a framework for problem elicitation and checking, propagation and learning. I will discuss its construction and illustrate its properties using a number of simple examples.
Biography: Jim Smith is a Professor of Statistics at the University of Warwick and is Chairman of RISCU (the Risk Initiative and Statistical Consultancy Unit) there. He has particular interests in problems on the interface between Statistics and AI. In particular he has contributed to the development of the theory and application of Bayesian Networks and other graphical models, especially ones which express a dynamic or causal environment.
Dr Tao Xiang