Machine Learning II (2007/2008)

This course covers important material left out in the Gatsby Probabilistic and Unsupervised Learning course, and is targeted at students interested in machine learning research. The aim of the course is to impart both a broader knowledge of the field, as well as in-depth knowledge of current advanced Bayesian techniques as used in machine learning.

The course is run as a number of independent modules covering a range of topics. Some modules are run as taught lectures, others seminar style (presented by students), or as coding days (morning lecture followed by a day of implementing the algorithm).

The course is run primarily for new Gatsby machine learning students for whom it is mandatory. Students, postdocs and faculty from both within and outside the unit are welcome to attend, but should contact Yee Whye in advance.

A good background in statistics, calculus, linear algebra, and computer science, as well as familiarity with the material in the Probabilistic and Unsupervised Learning course. You should thoroughly review the maths in the following cribsheet [pdf] [ps] before the start of the course. Knowledge of Matlab or Octave will be required for the coding days.

There is no required textbook. However, the following are excellent introductory texts for many of the topics covered here.

David J.C. MacKay (2003) Information Theory, Inference, and Learning Algorithms, Cambridge University Press. (also available online).

Christopher M. Bishop (2006) Pattern Recognition and Machine Learning, Springer-Verlag.

Stephen Boyd and Lieven Vandenberghe (2004) Convex Optimization, Cambridge University Press. (also available online).

Bernhard Scholkopf and Alex Smola (2002) Learning with Kernels, MIT Press.