Highlights
May 2009 An extended technical report on the Sticky HDP-HMM, with applications to speaker diarization.
May 2009 I'm co-editing a SP Magazine special issue on Graphical Models in Signal Processing.
November 2008 Two nonparametric Bayes papers at NIPS 2008: shared segmentation of natural scenes using dependent Pitman-Yor processes, and time series modeling with the HDP switching LDS.
May 2008 Release of the TDP Matlab toolbox for part-based, nonparametric modeling of objects and scenes. See our IJCV paper.
Erik B. Sudderth
Postdoctoral Scholar
Electrical Engineering & Computer Science
University of California, Berkeley
I am a postdoc at UC Berkeley, where I work with Professors Michael Jordan and Stuart Russell. My research interests span topics traditionally studied in statistics, machine learning, computer vision, and signal processing. Much of my recent work has explored vision systems which segment, recognize, and track objects in complex natural scenes. I believe data-driven, nonparametric Bayesian statistical methods provide a very promising framework to address these problems. My more abstract statistical research is typically inspired by the practical challenges of learning from large, richly structured datasets.
In June of 2006, I completed my Ph.D. in the EECS department at MIT, where I was advised by Professors Alan Willsky and William Freeman. The background chapter of my thesis provides a tutorial introduction to statistical machine learning, including probabilistic graphical models; Monte Carlo and variational inference algorithms such as belief propagation; and nonparametric Bayesian methods based on the Dirichlet process.
Moving to Brown University
In the summer of 2009, I will join the Brown University Department of Computer Science as an Assistant Professor. Brown provides an exciting, interdisciplinary environment for research in machine learning and computer vision:
- Brown Center for Vision Research
- Brown Machine Learning Reading Group
- Machine learning research at Brown: Artificial Intelligence, Robotics, Pattern Theory, Natural Language Processing
Contact Information
University of California, Berkeley
EECS Department
527 Soda Hall
Berkeley, CA 94720-1776
Tel: (510) 642-9582
