Please send further topic suggestions, and papers to read to: szummer at
microsoft dot com.
We encourage two leaders per discussion topic if you can!
[Names in square brackets] - indicate people
interested in leading discussions on the topic. More is better!
The order of topics is to be determined.
CVPR 2004 papers
ICML 2004 papers
Michael Mitzenmacher (Harvard University) papers [David MacKay]
Nonparametric Bayesian Methods
- e.g. NIPS workshop background papers (mailed out)
- for language models (Radford Neal)
[Phil] [Hugo][Martin]
Language modelling -
Roni Rosenfeld's tutorial: energy-based models [Phil]
Maximum likelihood training of MRFs – overview of huge literature
[Andrew] [Tom] [Markus]
Observable operator models for time series modelling (Jaeger) – generalization of HMM
Multilabel / Hierarchical classification techniques [Martin][John]
Frequentist methods (tutorial)
- for nonparametric models/tests (e.g. Kolmogorov / Smirnov test) [Ralf]
- false discovery rate
- frequentist test theory
External speaker?
Machine learning for games [Thore] [Ralf] [David Stern]
replace search with learning
. Baum – Bayesian tree search
Reinforcement learning
Information bottleneck [Markus]
Search algorithm
Applications
Compilers that learn to optimize (COLO) [Markus]
Learning techniques for detecting bugs in programs (Zheng NIPS03)
Amos Storkey - scheduled to come
Frequentist statistician
Martin Wainwright [Tom][Martin]
Radford Neal [David]
David Spiegelhalter [John]
Tali Tishbhy
DJVU
Overview [David]
Arithmetic coding [Phil]
Image modelling [John] [Martin]
[David x 2]
CRF
– (Herbert & Kumar)
[Chris][Martin][Hanna]
Variational Methods :
Wainwright and Jordan tech report – 2 sessions [Ralf] [John][Tom][Chris]
Inference in MRFs [Vladimir]