|
|
 |
|
|
|
Ulrike von Luxburg, Dr.
Senior Research Scientist |
|
|
|
|
|
|
|
MPI for Biological Cybernetics
Dept. Schölkopf
Spemannstraße 38
72076 Tübingen
|
| Telephone: | |
+49-7071-601 540 |
| Telefax: | |
+49-7071-601 552 |
| Room: | |
206 |
| e-mail: | |
ulrike.luxburg@tuebingen.mpg.de |
|
|
|
|
|
|
|
| |
List of Publications is
here
I am co-organizing the
NIPS Workshop "Clustering: Science or Art?". For information about the workshop and the call for contributions see
here.
Scientific interests
In a broad sense, my field of research is the theoretical analysis of machine learning algorithms. More particular, I am currently working on two major topics:
Theoretical foundations of clustering: Given that clustering is one of the most popular techniques for exploratory data analysis, it is intriguing to see how little is known about theoretical aspects of clustering. For example, for most clustering algorithms consistency statements do not exist, and we are far from being able to give performance guarantees or confidence statements on their outcomes.
My second area of interest is the combination of graph theory with machine learning and statistics. My goal is to study the statistical properties of graph based machine learning algorithms, for example in order to answer questions such as:
How should we construct the similarity graphs in graph based learning algorithms? Which properties of graphs are attractive for machine learning?
Which ones are misleading?
My research won several
awards :-)
GraphDemo
We (that is,
Matthias Hein
and me) wrote a GraphDemo package in Matlab. It provides Matlab GUIs to explore similarity graphs and
their use in machine learning. Essentially, it is a click and play interface to highlight the behavior of
different kinds of similarity graphs and to demonstrate their
influence on the outcome of machine learning algorithms (such as spectral clustering and semi-supervised learning). Primarily it has been written for teaching purposes, but we found it quite illuminating ourselves. For more information and download see the
GraphDemo main page
Teaching
Winter term 2007/2008: Lecture course on "Mathematical theory of machine learning"
at the
Mathematics Department of the University of Tübingen
Summer 2007:
Course on "Clustering - Theory and Algorithms" at the
Pascal Machine Learning Bootcamp, Barcelona, Spain
Summer 2007:
Practical Sessions on "Spectral Clustering and other graph based algorithms" at the
Machine Learning Summer School,
Tübingen, Germany
Summer term 2006: Lecture course
"Unsupervised Learning" , TU Darmstadt
Winter term 2005/06: Seminar
Machine Learning with Graphs, TU Darmstadt
Diploma or masters theses: I have several topics for diploma theses or master theses, both for maths and computer science students. If you are interested, please send me an email.
Workshops I (co)organized:
July 2007: Workshop on
Stability and Resampling Methods for Clustering
December 2005:
NIPS Workshop on Theoretical Foundations of Clustering
|
|
|
|
|
|
|
|
