|
|
Michael Jordan, University of California-Berkeley
|
|
|
|
Hierarchical Nonparametric Bayesian Models |
|
| Abstract |
Research on graphical models has tended to concentrate on exponential
families of probability distributions. This focus on parametric families
leads to model selection problems, and unfortunately many standard model
selection procedures are based on assumptions that are not well tailored
to the graphical model setting. An alternative is to consider
nonparametric Bayesian models; these are models in which the priors are
general stochastic processes. Mimicking the hierarchical modeling
concepts that are so useful in graphical models, we are led to study
hierarchical nonparametric Bayesian models. I discuss hierarchies of Dirichlet processes and hierarchies of beta processes; these are nonparametric priors that yield interesting "combinatorial" forms of statistical shrinkage. [Joint work with Yee Whye Teh and Romain Thibaux.] |
|
About the Speaker |
Michael Jordan is Professor in the Department of Electrical Engineering and Computer Science and the Department of Statistics at the University of California at Berkeley. He received his Masters from Arizona State University, and earned his PhD from the University of California, San Diego. He was a professor at the Massachusetts Institute of Technology for eleven years. He has published over 250 research articles on topics in statistics, computer science, electrical engineering, cognitive science and computational biology. His research in recent years has focused on probabilistic graphical models, kernel methods, nonparametric Bayesian methods and applications to problems in bioinformatics, information retrieval, and signal processing. He is a Fellow of the American Association for the Advancement of Science, the IEEE, the IMS and the AAAI. He was named an Institute Medallion Lecturer by the IMS in 2004. |
|
