IC Mondays seminars - Bayesian Machine Learning: Theory, Algorithms, and Large Scale Applications
Event details
| Date | 26.04.2010 |
| Hour | 16:15 |
| Speaker | Dr. Matthias Seeger, University of Saarland |
| Location |
INM 202
|
| Category | Conferences - Seminars |
Abstract
Today's real world problems of information processing rely on inference and decision making from uncertain knowledge. Modern science and medicine need autonomous tools for acquisition and sifting of data or experimental planning. Bayesian graphical modelling poses hard computational challenges in practice. Some of these have successfully been addressed in machine learning, calling on ideas from stochastic simulation, convex optimization, numerical mathematics, and graph theory.
I will give an overview of my work, alongside many international coauthors, on progress in probabilistic machine learning. My contributions range from novel theoretical tools to simplify and sharpen analysis of nonparametric Bayesian methods, over insights and development of variational approximate inference techniques, to recent milestones in convex relaxations and scalable algorithms for image reconstruction models. With the latter, Bayesian computations can be
performed over full high-resolution image bitmaps for the first time. I will demonstrate how magnetic resonance imaging acquisition patterns can be autonomously optimized driven by this novel Bayesian framework, resulting in substantially shorter scan times.
Biography
Matthias Seeger received his CS Diploma degree from Karlsruhe university in 1999 (distinction), his PhD from Edinburgh university in 2003 (with Christopher Williams). He did postdoctoral research at the University of California, Berkeley (with Michael Jordan, Peter Bartlett) and at the Max Planck Institute, Tuebingen (with Bernhard Schoelkopf). At present, he leads an independent research group at the Max Planck Institute and Saarland University, Saarbruecken, supervising three PhD students. He works on the theory, algorithmics and scalable implementation of Bayesian technology and probabilistic machine learning. He is the author of numerous journal and international conference publications, organized workshops at and served in senior program committees of leading international machine learning conferences (NIPS AA area chair 2004, 2010; UAI 2009; AISTATS 2010).
Links
Practical information
- General public
- Free
Contact
- G.Rochat