MSc in Intelligent Systems 2007-2008
Advanced Topics in Machine Learning (part II)


Overview

In this course we will discuss linear probabilistic models for regression, Gaussian processes, linear and nonlinear dynamical systems (including sigma points filters and particle filters). We will also cover Dirichlet processes and continuous-time stochastic processes (including the basics of Ito calculus and an introduction to stochastic differential equations) .

The main objective of the course is to study a number of advanced topics in Machine Learning in detail and to implement two techniques related to the topics discussed in class. Several guest speakers will also be invited to talk about their respective field of expertise.

For additional details on the course, please refer to the syllabus.

Unless otherwise stated, the course will take place on Fridays from 10:00 to 13:00 in room 339 of the Rockefeller building, University College London.

For more information about the course or if you want to attend (and are not a MSc student), please contact Cedric Archambeau.

Schedule

11/01 Linear models for regression and Gaussian processes
  Introduction and basic concepts.
Probabilistic linear models for regression.
Gaussian processes for regression.
   
18/01 Filtering and smoothing in dynamical systems
  Hidden Markov models and linear/nonlinear state space models.
(Kalman, extended Kalman and sigma points Kalman filters/smoothers; particle filters)

Guest speaker at 12:00: Frank Wood (Gatsby unit).
   
25/01 Dirichlet processes, infinite mixtures and extensions
  Dirichlet processes and DP mixtures.
Hierarchical Dirichlet processes and DP mixtures of regressors.

Guest speaker at 12:00: Yee Whye Teh (Gatsby unit).
   
05/02 (Tuesday!) Lab session (14:00 - 17:00, Malet Place Engineering building, room 1.04!!!)
  DP mixtures of linear dynamical systems.

Guest speaker at 15:00: David Barber (CSML).
   
07/02 (Thursday!) Continuous-time stochastic processes (13:30 - 16:30, Malet Place Engineering building, room 6.25!!!)
  Introduction to stochastic differential equations and Ito calculus.
Continuous-time stochastic processes.
Wiener process, Diffusion processes and Levy processes.
Variational inference in Diffusion processes.
   

Last updated on 07 February 2008.