University of Bath

Dr Joanna Bryson

Dr. Bryson supervises two sorts of projects: AI projects, primarily those to do with making the design of AI systems easier, and NI projects, primarily those modelling the evolution of social and cognitive behaviour. NI projects normally include a systems AI component, since the tools we use for modelling natural intelligence are all AI. Improving the AI or tools is one way to make the computer science component of the project clear. Being your own user is also the best way to make sure your tools are actually useful.

Similarly, for the AI projects, even if improving a tool is their main thrust, you should still have a science or game development project to conduct as well, since developers don't tend to really improve tools they aren't actively using. See the AmonI software page for details of some of our existing projects, and related papers on the designing intelligent systems page.

Dr. Bryson meets with all her students once a week in a group meeting, so her students can also experience peer supervising and supervision. There will of course also be a few longer individual meetings at critical periods in both terms. She is not on campus until late August, but can be contacted by email to negotiate projects. She sometimes accepts novel project proposals from strong students, and is happy to negotiate about and adjust her own proposals to the interests and capabilities of a particular student.

Examples of possible NI projects this year include:

Modelling matriarchical dominance structures non-human contexts

Modelling the social structure of lemurs or spotted hyenas. This would be interesting in itself, but particularly interesting if we could understand why these species have evolved to have female dominant social structures.

Modelling fission / fusion social behaviour in non-human species

Modelling the fission/fusion society and inter-troop relations of chimpanzees, spider monkeys or Mongolian Asses. Some work on this was done by Rob Jenks in 2008/9, but it needs to be extended so that it can be evaluated against animal data.

Modelling cultural variation

Modelling cultural variation in the use of anti-social punishment. See "Antisocial Punishment Across Societies" Science 319, 7 March 2008, 1362-1367 by Benedikt Herrmann, Christian Thöni and Simon Gächter. [abstract], [article], [supplementary materials], [commentary by Herb Gintis], [further information].

For previous projects that have been related to this work, see Christos Bechlivanidis, Avri Bilovich, and Ando Yasushi's dissertations on my previous dissertations page, and papers with Hagen Lehmann and Ivana Cace on my publications page. Hopefully Rob Jenks' dissertation will also soon be on the dissertations page.

Examples of possible AI projects for this year:

Adapting a Video Game to do Psychological Experiments

This project would involve conducting experiments in human implicit task learning through looking at how they perform in navigation and exploration tasks in a video game. You would need to modify a game (probably Unreal Tournament) to include three customized levels to correspond to the three phases of the experiment, and also to record statistics on the subjects, for example their route choices and their reaction times when faced with decisions. You would also need to conduct the experiments and write up the results formally.

Visualization and Maintenance of an Intelligent Virtual Agent

In a recently completed PhD project, Emmanul Tanguy built the Emotionally Expressive Facial Animation System (EE-FAS). The basic system is very modular, having been built on top of psyclone. However, understanding the architecture is not trivial for new users. This project would involve making it easier to see what EE-FAS is doing, and possibly to maniuplate the system. One way to do this might be to make a general GUI tool for enhancing the usability of psyclone. Another would be to focus on the emotion model of EE-FAS, showing their levels and interactions, allowing the interactions to be adjusted, or even for new basic emotions to be added. A truly exceptional project might do all these things.

The Behavior-Oriented Design of Online World Avatars.

Behavior-Oriented Design (BOD) is an iterative methodology for building artificial intelligence. It is based on both object oriented design, dynamic AI plans and somewhat like extreme programming. Previously it has been applied to robots, scientific simulations an unreal-tournament characters. We would like to extend this to either on-line games or VR social spaces, such as Second Life. The first part of the project would be evaluating games or realities for their accessibility & appropriateness to AI techniques like BOD.

Last modified: Monday, 9 August 2010, 07:15 AM