EE Department Intranet -
Close window CTRL+W

COMP70019 Probabilistic Inference

Lecturer(s): Dr Mark van der Wilk


For up to date course information, see the DoC website:
493;Intelligent Data and Probabilistic Inference;;i4

The course is concerned with probabilistic methods for modelling data and making inferences from it. The first part of the course introduces Bayesian Inference and Networks and includes probability propagation and inference in singly connected networks, generating networks from data, and calculating the network accuracy. The course then goes on to consider highly dependent data and special techniques for exact and approximate inference in these networks. The next topic to be covered is data modelling using distributions and mixture models. The topic of sampling and re-sampling is covered along with data reduction by principal component analysis and special problems that occur with small sample sizes. The last part of the course is concerned with classification using Linear Discriminant analysis and Support Vectors.

The emphasis of the course is algorithmic rather than mathematical, and the coursework is a practical programming exercise in analysing data from a study into the prognosis of Hepatitis C.

Note that the course does not include non-probabilistic methods of data analysis such as Neural Networks, Fuzzy Logic or expert systems.

Learning Outcomes


Exam Duration: 2:00hrs
Exam contribution: 80%
Coursework contribution: 20%

Term: Spring

Closed or Open Book (end of year exam): Closed

Coursework Requirement:
         To be announced

Oral Exam Required (as final assessment): N/A

Prerequisite module(s): None required

Course Homepage:

Book List: