i++ Departmental Newsletter

Week commencing 18 February 2008

Previous Newsletters

IEEE to Launch New Journal

A proposal for a new journal has just been accepted by the IEEE Board of Directors.  The journal is IEEE Transactions on Computational Intelligence and AI in Games.
 
This will be a fantastic boost for research in the area worldwide, and underlines the importance and intellectual challenge of the field.
 
Dr Simon Lucas submitted the journal proposal in July 2006 in consultation with the Games Technical Committee, and with the strong support of the IEEE Computational Intelligence Society, in particular Dr David Fogel (Natural Selection inc., and IEEE CIS President), Professor Jim Keller (University of Missouri, IEEE CIS VP Publications), and Professor Vincenzo Piuri (University of Milan, Former IEEE CIS President).
 
Establishing a new IEEE Journal is a major undertaking that requires approval by a number of committees at a number of stages.  In this case it was wonderful to see the proposal receive approval first time at every
stage.  A steering committee will now begin the selection process for an editor in chief, after which the editorial board will be appointed, with the first issue of the journal due to be published in March 2009. 
 
The scope of the journal appears below.
  
The IEEE TRANSACTIONS ON COMPUTATIONAL INTELLIGENCE AND AI in GAMES (T-CIAIG), published four times a year, publishes archival journal quality original papers in computational intelligence and related areas in artificial intelligence applied to games, including but not limited to video games, mathematical games, human-computer interactions in games, and games involving physical objects.  Emphasis will also be placed on the use of these methods to improve performance in and understanding of the dynamics of games, as well as gaining insight into the properties of the methods as applied to games.  It will also include using games as a platform for building intelligent embedded agents for the real world.  Papers connecting games to all areas of computational intelligence and traditional AI will be considered.

Essex Games Research Makes Top 10

The research work on automated content creation for games by Julian Togelius,  Renzo de Nardi, and Simon Lucas has been recognised in a prestigious list of worldwide research findings.  This list is presented as part of an annual talk given at the Game Developers’ Conference regarding the top 10 research findings worldwide in games studies for the previous year.  For an overview of what was said visit Raph Koster's website or the official list with reference can be downloaded form the authors’ web site at avantgame. These are very well-respected game studies researchers, and making it on to the list is an impressive achievement.

The research dealt in particular with using machine learning methods to build an avatar that drives like a particular human player, and then automatically evolving car tracks that offer that avatar (and hence the player) an appropriate degree of interest and challenge.  Some sample tracks are shown below.

CES Research featured on ITV Local Website

Earlier this year Anglia TV visited to shoot some video  footage for the ITV local web site. These videos are now available.  Click on the images below to see Professor Hu talking about intelligent wheelchairs,  and Dr Lucas discussing the machine learning challenge of playing  Ms Pac-Man.

Forthcoming Seminar

Dr. Damien Coyle, Ulster University             Email: dh.coyle@ulster.ac.uk

(Suggested by Francisco Sepulveda)  

Friday February 22, 3.00pm, Room 1N1.4.1

Advancing Brain-Computer Interface Technology with a Prediction-based Preprocessing Framework

 Abstract – Brain-Computer Interface (BCI) research is growing at a significant pace and, since the beginning of the 21st century, has seen explosive growth. The depth and breadth of BCI research in progress today is indicative of its application potential. BCI technology can provide a communication pathway from the brain to the computer which does not rely on neuromuscular control therefore there are many potential beneficiaries of the technology. These include people with neuromuscular deficiencies due to disease or spinal chord injury. Being able to offer these people an alternative means of communication through BCI could have an obvious impact on the quality of life of these people. There are other applications of BCI, yet to be fully proven and exploited, such as neurofeedback for stroke rehabilitation and epileptic seizure prediction. BCI is also emerging as an augmentative technology in computer games and for alternative computer interaction in virtual or real environments technology.

As yet BCI is a nascent technology. There have been many advances but there are still a significant number of problems and issues to be resolved. To date, the majority of Dr. Coyle’s research has been focused on the application of computational intelligence (CI) to tackle complex biosignal analysis and discrimination problems for non-invasive electroencephalogram (EEG)-based BCI technology. BCI requires accuracy, speed and autonomous adaptation capabilities and Dr. Coyle has attempted to address these requirements by developing a unique preprocessing framework, referred to as neural-time-series-prediction-preprocessing (NTSPP). This framework permits multiple-step-ahead prediction of the EEG time-series, where different regression/prediction models are trained to specialize in predicting different EEG signals. Time- and frequency-domain features are extracted from the predicted signals and classified. Due to the specialisation of each regression model on the type of data on which it was trained to predict, feature separability (i.e., discriminability) and inter-session (i.e., different day) generalisation is improved by the NTSPP framework. NTSPP maps the original signals to a higher dimensional space where the Euclidean distance between class means is increased and the inter-class correlation and the intra-class variance is reduced. Multiple tests on multiple subjects, offline and online, have shown that the NTSPP framework improves classification accuracy (CA), information transfer (IT) rate, performance at the early stages of BCI usage and has the potential to contribute to better parameterless and autonomous system adaptation. These performance criteria are critically important for successful application of BCI technology. Dr. Coyle’s presentation will describe the NTSPP framework in detail and how it has been applied in BCI. Many of the problems which confront BCI researchers/developers will be illustrated and possible ways of tackling these problems along with an outlook for the future.