Courses

The course covers some advanced models in machine learning. 
The models are analyzed mainly from a Bayesian perspective. 
The course is also a master's course on the LiU programme 'Statistics and Machine Learning'.

The course is organized into four topics:
- Graphical Models
- Hidden Markov Models
- Gaussian Process Regression and Classification
- State-space models

Each topic includes lectures, a computer lab and a follow-up seminar.

There will be three 3-days gatherings each comprising 18h of lecturing. The first meeting will take place in Växjö, 12-14 October, the second meeting will be held in Stockholm, 5-7 December and the third one is organized in Uppsala, 8-10 February.

Instructors who so far have agreed to participate are Thomas Holgersson, Björn Holmquist, Hans Nyquist, Dietrich von Rosen, Rolf Sundberg and Silvelyn Zwanzig. There will be another three so each meeting will be run by three instructors. 

Content

In conjunction with the LinStat 2014 a mini-course (8h) entitled 'Survey sampling and linear models' will be held by professor Stephen Haslett, Massey University, New Zealand and professor Simo Puntanen, University of Tampere, Finland, August 23-24.

The course is sponsored by GRAPES and GRAPES will refund travel and accommodation costs for PhD students at Swedish universities.

In this course we introduce causal inference with main focus on two approaches; i) the potential outcome framework/ Rubin Causal Model and ii) Causal inference with graphical models.
Identification of causal effects in the two approaches is studied. Non and semi-parametric estimators of causal effects commonly associated with the potential outcome framework will be presented. A general  introduction to graphical models is also given.