We will largely use the book Bayesian Reasoning and Machine Learning by David Barber (Cambridge University Press, 2012), together with additional material as needed. A free online version of the book is available here.

Broadly speaking, the course covers four topics:

  • Probabilistic graphical models
  • Exact inference
  • Learning
  • Approximate inference and learning

Probabilistic graphical models will be treated in more detail than in Barber’s book.

To pass the course you must understand the material on the slides, the tutorials, and the material listed below as “required” (unless marked as not examinable). However, you may understand the content better if you work through the optional material too.

Programme

Week Lecture Tutorial
1 Tue Jan 16 Slides: Required: Optional:
Fri Jan 19 Slides: Required: Optional:
2 Tue Jan 23 Slides: Required: Optional:
Fri Jan 26 Slides: Required: Optional:
3 Tue Jan 30 Slides: Required:
  • Barber: Sections 4.1, 4.2 till 4.2.1
Optional:
Tutorial 1:
Fri Feb 02 Slides: Required: Optional:
4 Tue Feb 06 Slides: Required:
  • Barber: Section 4.5
Tutorial 2:
Fri Feb 09 Slides: Required: Optional:
5 Tue Feb 13 Slides: Required: Optional: Tutorial 3:
Fri Feb 16 Slides: Required: Optional:
No class or tutorials the week of Feb 19
6 Tue Feb 27 Slides: Required: Optional: Tutorial 4:
Q&A for assignment
Fri Mar 02 Slides: Required: Optional:
7 Tue Mar 06 Slides:
  • The Learning Problem (TBC)
Tutorial 5:
Fri Mar 09 Slides:
  • Learning from Fully Observed Data: Undirected Graphical Models (TBC)
8 Tue Mar 13 Slides:
  • Learning from Fully Observed Data: Undirected Graphical Models (TBC)
Tutorial 6:
Fri Mar 16 Slides:
  • Approximate Inference (Monte Carlo and variational) (TBC)
9 Tue Mar 20 Slides:
  • Approximate Inference (Monte Carlo and variational) (TBC)
Tutorial 7:
Fri Mar 23 Slides:
  • Learning from Partially Observed Data (incl latent variable models) (TBC)
10 Tue Mar 27 Slides:
  • Learning from Partially Observed Data (incl latent variable models) (TBC)
Tutorial 8: Q&A:
Fri Mar 30 Slides:
  • Learning from Partially Observed Data (incl latent variable models) (TBC)
11 Tue Apr 03 TBD
Fri Apr 06 TBD