Matthias Seeger, Suvrit Sra, John Cunningham
International Conference on Machine Learning 26 (2009).
With this workshop, we brought together experts from numerical mathematics
and machine learning. While many publications and most public software in
machine learning fall short of sound numerical practice, thorough numerical
analysis of most machine learning algorithms is probably too much to ask for.
We stressed the importance of starting to think about how to bridge this gap
in a tractable manner.
[homepage, videos]
Mattias Seeger, David Barber, Neil Lawrence, Onno Zoeter
Neural Information Processing Systems 20 (2007).
Many of the most important problems in Machine Learning and related application
areas are most naturally and succinctly treated using continuous variable
models. Yet most research work is done for discrete variable inference. With
this workshop, we aimed to assess the status quo for continuous variable
deterministic (variational) inference techniques, and the nature of major
similarities and differences to the discrete variable world.
[homepage, videos]