Seminar Details
| Date |
31-3-2004 |
| Time |
16:00 |
| Room/Location |
Sala conferenze, 322, 3 piano |
| Title |
Statistical inverse problems and hierarchical models |
| Speaker |
Professor Erkki Somersalo |
| Affiliation |
Department of Mathematics Technical University of Helsinki Helsinki, Finland |
| Link |
http://www.cwru.edu/
|
| Abstract |
We consider inverse problems from the point of view of Bayesian statistics,
in which the problem is a statistical probelm of inference based on
indirect observations. Often,the prior density depends on parameters
that are badly known. The selection of these parameters is sometimes
tantamount to selecting regularization parameters in classical inversion
schemes. It is shown how these parameters are determined witihn the
Bayesian framework by using hierarchical models.
The method is applied to the MEG problem of localizing source currents.
|
|
|
 |
