Outline:

Bayesian inferential reasoning weighs evidence on a real valued scale and chooses a decision that maximises the expectation of a real valued utility function. This is a natural way of incorporating scientific evidence of various kinds in a transparent way to inform decisions. On the other hand a different way of reasoning is based on argumentation theory and is widely applied to situations like those found in a court of law. Here a rhetorical framework is built where proponents for and against a decision make arguments for and against the evidence supporting the different acts. A third party weights the strength of these arguments to determine the best course of action.

This project will examine the formal mathematical background that supports these different methods. The student and supervisors one an expert in Bayesian reasoning and the other in argumentation theory will then develop ways of harmonising these methods to leverage the advantages of both into a single system of reasoning. The method will be tested against software called Open Clinic – developed by the second supervisor - which provides decision support medical clinicians. Background information on these two systems can be found in Smith, J.Q. (2010) “Bayesian Decision Analysis” Cambridge University Press and Fox, J and Das S.(2000)”Safe and Sound” MIT Press.

 

Prerequisites:

The ideal student would to have either a four year maths and computing or maths and statistics or a three year degree in the same with an MSc in Statistics or Machine Learning. The student will interested in both developing new methodology and applying this within an established decision support system.