PhD thesis on PAC-Bayesian bounds and sparse Gaussian processes

[mseeger@EECS.Berkeley.EDU]

Dear colleagues,

my PhD thesis is available online at
  www.dai.ed.ac.uk/~seeger/papers/thesis.html.

It mainly deals with:
- PAC-Bayesian generalisation error bounds and
  applications to Gaussian process classification
- Sparse approximations for linear-time inference in
  Gaussian process models

Please find abstract and table of contents on the
website.

You might also be interested in the tutorial paper
  Gaussian Processes for Machine Learning,
available at
  www.dai.ed.ac.uk/~seeger/papers/bayesgp-tut.html
which is extracted from the thesis, but is
self-contained. An abstract follows.

Best wishes, Matthias.

----

Gaussian Processes for Machine Learning

Gaussian processes (GPs) are natural generalisations of multivariate
Gaussian random variables to infinite (countably or continuous) index
sets. GPs have been applied in a large number of fields to a diverse
range of ends, and very many deep theoretical analyses of various
properties are available. This paper gives an introduction to Gaussian
processes on a fairly elementary level with special emphasis on
characteristics relevant in machine learning. It draws explicit
connections to branches such as spline smoothing models and support
vector machines in which similar ideas have been investigated.

---
Matthias Seeger              Tel:
485 Soda Hall, UC Berkeley   Fax: 510-642-5775
Berkeley, CA 94720-1776      www.dai.ed.ac.uk/~seeger