i++ School Newsletter

Week commencing 12 July 2010

Previous Newsletters

£400k research grant awarded to CSee Academics

Professor Henson, Dean of International Development, has recently secured a £400k research grant to undertake joint work between Professors Vic Callaghan and Hani Hagras at Essex and Dr Daniyal Al Ghazzawi, Head of Information Systems, Faculty of Computing and Information Technology, King Abdul Aziz University (KAU), Jeddah. The project aims to investigate what is necessary to scale-up existing technology in intelligent inhabited environments, from smaller facilities such as the Essex intelligent apartment, iSpace, to multi-story, large footprint buildings. Professor Henson said, "Our relationship with KAU is long-standing. What began through student recruitment visits some years ago, later blossomed in to income-generating capacity-building activities concerning curriculum and staff development, new PhD programmes in Computer Science, and a range of educational and research connections in Biological Science. Most recently we spotted an opportunity to advise the Faculty of Computer and Information Technology on its research ambitions in Intelligent Systems. The new research project is, in essence, just a preliminary to a much bigger project in Jeddah that, if successful, aims to build a new research centre in Intelligent Systems that would itself, as an intelligent building, be a subject of investigation."  

Position for a Senior Research Officer investigating Monte Carlo Tree Search

Senior Research Officer

Monte Carlo Tree Search – Application to Video Games and Real-Time Control

(Ref.RE134) Artificial Intelligence (AI) research and the development of the multi-billion dollar video games industry have gone hand in hand for many years. Video games are by far the most prevalent way that the public encounter AI techniques on a day to day basis, and the desire for better video games has driven AI research in areas such as move/path planning, decision making, non-player character (NPC) behaviour and the automated generation of game content. A recent development of Monte Carlo methods called the Upper Confidence Bounds for Trees (UCT) method promises to have a profound impact on AI for games. Applications of Monte Carlo Tree Search (MCTS) are not limited to games and have potential benefits for almost any domain where simulation and statistical modelling can be used to forecast outcomes, such as planning, decision support, economic modelling, behavioural analysis, and so on.

The proposed research will develop and evaluate novel extensions of the MCTS to increase its applicability to a broad range of game-related domains including, with the Essex part of the project focussing on its use for move planning and decision making in infinite, continuous real-time environments such as video games.  In particular, the project will explore the following topics within the wider context of MCTS: approximate modelling, real-time MCTS, multi-objective MCTS, learning within MCTS, and overall optimisation of MCTS agents.

We have received substantial funding from EPSRC to investigate the full potential of Monte Carlo Tree Search, in collaboration between the University of Essex, Imperial College London, the University of Bradford, AI Factory Ltd., Introversion Software Ltd. and Nestorgames Ltd.

For more details see here.

The Post  

You will work closely with Professor Simon M Lucas and the Game Intelligence Group at the University of Essex, and also with the other project partners.  The appointee will be expected to design algorithms, write software, conduct experiments, present their work to other consortium members and at international scientific conferences, and write high-quality journal and conference papers.

A First class or 2.1 honours degree or equivalent in a scientific discipline with substantial experience of computer programming and discrete mathematics is essential for this post.  A PhD in computer science / artificial intelligence is also expected.

Excellent English language skills are required and candidates whose first language is not English need to have IELTS 7.0 or equivalent.  An established or developing track record of research and publications is necessary.

Good technical programming skills are required, preferably in a number of languages such as Java, C#, C++, Python etc.  Candidates should have a sound knowledge of artificial intelligence, discrete mathematics and the design of algorithms and data structures.  The ability to write GPU programs is and an enthusiasm for games is also desirable.

This post is fixed-term for 3 years to commence from 1 October or as soon after by agreement

The University and Location

The University of Essex is one of the leading research-oriented universities in the UK, consistently finishing in the top ten UK universities in the league tables for research.  The studentship is based at the University’s Colchester Campus set in Wivenhoe Park, occupying 200 acres of Constable Country.  Colchester is Britain’s oldest recorded town, and has excellent transport links with London just 50 minutes away by train, and Stansted airport offering a multitude of low-cost flights to Europe and beyond.  Colchester is situated in East Anglia, one of the sunniest parts of the UK.

Salary

£29,853-£31,671 per annum                   

Closing Date

16 August 2010

Apply online.  If you have a disability and would like information in a different format telephone (01206) 873521/874588.

University of Essex Jobs site

CEEC 10 - Submission Deadline Extended

The paper submission deadline for CEEC 10 has been extended to 23 July. See the website for further information.

staff news

Professor Richard Bartle

Professor Richard Bartle gave the opening commentary at the prestigious Computer Games Online convention in Leipzig, Germany. Half an hour before Germany played Spain in the semi-finals of the World Cup, he had to explain to an audience of politicians, journalists and industry personnel why being at the convention was more important than sitting in front of their wide-screen TV at home, clutching a beer. He failed, but so did the German team.

Next day, he was interviewed by three TV stations, two radio stations and a newspaper.

Paper Published

The following paper will be appearing in Electronics Letters as a featured article;

Shoaib Ehsan, Nadia Kanwal, Adrian F. Clark and Klaus D. McDonald-Maier, Improved repeatability measures for evaluating performance of feature detectors, Electronics Letters, Volume 46, Issue 14, p.998–1000, 8 July 2010

Abstract - The most frequently employed measure for performance characterisation of local feature detectors is repeatability, but it has been observed that this does not necessarily mirror actual performance. Presented are improved repeatability formulations which correlate much better with the true performance of feature detectors. Comparative results for several state-of-the-art feature detectors are presented using these measures; it is found that Hessian-based detectors are generally superior at identifying features when images are subject to various geometric and photometric transformations.

Today's Seminar

Wednesday 14 July at 16.00, Room 1N.1.4.1

Learning Fuzzy Valuation Models

Speaker: Dr Adam Ghandar, University of Adelaide

Abstract - Multi-factor asset valuation models are commonly used in the growing area of quantitative portfolio management, for example see Fama, 1996.

Computational intelligence (CI) combines heuristic techniques with flexible problem representations such as fuzzy logic and neural networks. It is intuitively the case that the processes underlying market prices change over time. This has also been investigated in behavioural finance, for instance see Shleifer 2000. So, an adapting approach to modelling seems likely to be advantageous. This presentation will discuss an application of evolving fuzzy rules to implement adaptive decision models for financial portfolio management and gives an overview of some results from simulations that consider transaction costs, stock splits, survivorship bias and other issues.

Compared with regression, a "classical" technique in financial modelling, advantages obtained using computational intelligence include: modelling the changing effects of model input variables; using a non-linear solution representation; and, interpreting huge amounts of data automatically. Compared with other soft computing techniques, the evolving fuzzy approach discussed differs in interpretability of solutions to humans; a resulting flexibility in the extent models are learned automatically or augmented by users; and by defining a search space that comprises If-Then rules structured to correspond with possible natural language statements about an information set including technical analysis indicators and accounting and company information.