Matthew Urry
homepage:http://www.mth.kcl.ac.uk/~matturry/
search externally:   Google Scholar,   Springer,   CiteSeer,   Microsoft Academic Search,   Scirus ,   DBlife

Description

I am interested in Machine Learning. I am currently focused on

  • Gaussian Mixture Models where I am looking at clustering of Gaussian mixtures at optimum solutions
  • Gaussian Processes with a focus on applying Gaussian process regression to discrete spaces in particular learning functions defined on the nodes of graphs.

I have been looking at predicting the learning curves of the GP (mean square error as number of examples is increased) and hope to look at model mismatch and evolving graphs in the future.


Lecture:

demonstration video
flag Exact learning curves for Gaussian process regression on large random graphs
as author at  Video Journal of Machine Learning Abstracts - Volume 1,
255 views