I am a senior research scientist with
the Helsinki Institute of Information Technology.
We developed a client-server distributed, hierarchical information system based on
probability theory, see Next Generation
Information Search and the ALVIS project.
I also edited the
Open Source Search web site,
the Component Analysis web site,
and supported the
Automated Software Engineering Group
at NASA Ames Research Center in
building AutoBayes, a system for automated synthesis
of statistical algorithms.
Seen here is a shot from Osakwe Ates's (then 4-year-old) photography portfolio.
Recent research activity is as follows:
- Tutorial on Open Source, Distributed and P2P Information Retrieval given at European Summer School on Information Retrieval", ESSIR, August 2007.
- Lots of talks and things better documented at the ALVIS Publications page.
- Theory of random surfer models for the web: paper available on request.
- "Discrete Component Analysis", book chapter with Aleks Jakulin, available on request.
- Discrete Principal Components Analysis, tutorial
slides from PASCAL's Subspace, Latent Structure and Feature Selection techniques: Statistical and Optimisation perspectives Workshop,
23nd February, 2005. Full length paper at the MPCA site. A complete rewrite has been done, but not currently
available on line.
- Applying Discrete PCA in Data Analysis
(Uncertainty and AI, 2004)
- Is Multinomial PCA Multi-faceted Clustering or
Dimensionality Reduction?(AI and Statistics, 2003)
- Automatic Derivation of Statistical Algorithms: The EM Family and Beyond (early draft for NIPS 2002)
- Multi-faceted Learning for Web Taxonomies (Semantic Web Mining Workshop, 2002)
- Variational Extensions to EM and Multinomial PCA (ECML-2002)
A System for Generating Data Analysis Programs from Statistical Models
Lecture Course on Graphical Models at University of Helsinki
for Oct-Dec. 2005. Lectures in PDF are available as L1, L2, L3, and L4.
My Docentship Lecture given at the University of Helsinki, 6th February, 2006 (in PDF and in Postcript, 2up, gzipped).
Source code for Multinomial PCA available on request. Its GPL'd and runs on 2Gb of text
in a few days on your desktop. Comes with a custom browser to inspect results.
See MPCA Page and published on Freshmeat.NET.
Some older activity:
Dept. of Computer Science, PL 68, 00014, University of Helsinki, Finland.
Email: Wray.Buntine AT hiit.fi
Thu Aug 30 20:26:20 EST 2007