Research

|  Home Page   | All Publications   | Grants  | PhD background |   MSc background |

A full list of downloadable copies of my papers are available All here  

Data Mining

I am part of the Data Mining Group in the School of Computing Sciences and have been involved with data mining projects with Norwich Union, Master Foods and Centrica. My main current areas of interest are in time series data mining and super computer data mining. We have a senior research associate, Ian Whittley working on an EPSRC funded project developing super computer data mining tools and a PhD student, Mike Powell on an EPSRC CASE award with Master Foods, studying time series data mining.


Papers

 

·  A. J. Bagnall, G. Janakec, M. Powell A likelihood ratio distance measure for the similarity between the fourier transform of time series , accepted for publication the Ninth Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD-05)

A. J. Bagnall, G. Janakec, Clustering Time Series with Clipped Data, to appear in Machine Learning Journal, 2005

·  A. J. Bagnall, G. Janakec, Clustering time series from ARMA models with clipped data,SIGKDD, 10th International Conference on Knowledge Discovery in Data and Data Mining, 2004

·  A.A. Gill G. D. and Smith and A. J. Bagnall Improving decision tree performance through induction and cluster-based stratified sampling, Fifth International Conference on Intelligent Data Engineering and Automated Learning ,2004

·  A. J. Bagnall, G. Janakec, B. de la Iglesia and M. Zhang Clustering Time Series from Mixture Polynomial Models with Discretised Data in Proceedings of the second Australasian Data Mining Workshop, 2003

·  B. de la Iglesia, M. S. Philpott, A. J. Bagnall and V. J. Rayward-Smith Data mining rules using multi-objective evolutionary algorithms in Proceedings of the 2003 Congress on Evolutionary Computation, 2003

·  A. J. Bagnall, G. Janakec and M. Zhang Clustering Time Series from Mixture Polynomial Models with Discretised Data, Technical Report CMP-C03-17, School of Computing Sciences, University of East Anglia, September 2003. PDF

·  A. J. Bagnall. and G. C. Cawley Learning classifier systems for data mining : A comparison of XCS with other classifiers for the Forest Cover dataset, In Proceedings of the IEEE/INNS International Joint Conference on Artificial Neural Networks (IJCNN-2003), Portland, Oregon, USA, 20th-24th July, 2003. compressed postscript , PDF

Agents and Auctions

I have recently completed a Teaching Company Scheme (TCS are now Knowledge Transfer Partnerships ) with Effem Holdings, a division of Mars PLC. The objective of the project was to develop an electronic auction database and data mining analysis tool for data derived from online auctions. Business to business online reverse auctions are becoming popular as an alternative to requests for tender for purchasing a wide variety of raw materials. Large companies such as General Motors and Ford are conducting a significant proportion of their purchasing from suppliers via auctions. Several companies are involved with conducting these auctions, for example Freighttraders.com , Commerce One and Free Markets.
Some interesting background articles:
"Going, going, gone" by S. Tully in Fortune magazine,
"How I saved $100 million on the Web" by P. Judge in Fast Company,
"Dear supplier, this is going to hurt you more than it hurts me..." by B. Richards in eCompany now.
Despite their growing popularity, many people think online reverse auctions can damage business relationships (see here for example). I am also involved in a project with PhD student, Iain Toft, looking at adaptive agent architectures for learnig bidding strategies in auctions.

Papers

·  A. J. Bagnall A Multi-Agent Model of the UK Market in Electricity Generation , in Applications of Learning Classifier Systems, Studies in Fuzziness and Soft Computing, Vol. 150, 2004

·  A. J. Bagnall and I. Toft Autonomous Adaptive Agents for Sealed Bid Auctions",, workshop on Trading Agent Design and Analysis, part of international conference on autonomous agents and multiagent systems, 2004

·  A. J. Bagnall and I. Toft An Agent Model for First Price and Second Price Private Value Auctions , Lecture Notes in Computer Science, Vol 2936, 2004 PDF

Autonomous Adaptive Agents and Learning Classifier Systems

My PhD was on simulating the market in electricity generation using autonomous adaptive agents. The agents employ a hierarchical learning mechanism based on learning classifier systems (XCS in particular). I am supervising a PhD student (Joanna Zattuchna) who is developing new operators and structures for agents using XCS in maze environments. I am also looking at agents in simulated auction environments.


Papers

·  A. J. Bagnall and G. D. Smith, Game playing with autonomous adaptive agents in a simplified economic model of the UK market in electricity generation In Proceedings of IEEE-PES / CSEE International Conference on Power System Technology (Powercon 2000), Perth, Australia, 1st-5th December, 2000. compressed postscript, PDF

·  A. J. Bagnall, A Multi-Adaptive Agent Model of Generator Bidding in the UK Market in Electricity, In Proceedings of the AAAI Genetic and Evolutionary Computation Conference (GECCO-2000), pages 605-612, 2000, Morgan Kaufmann: San Francisco, CA, compressed postscript , PDF

Optimisation


I have a background in optimisation and am interested in applying optimization techniques developed at UEA to problems from auctions such as the winner determination problem for combinatorial auctions.


Papers

·  A. J. Bagnall, V. J. Rayward-Smith and I. M. Whittley, (2001), The Next Release Problem, Information and Software Technology, Vol. 43, 14, p. 883-890

·  A. J. Bagnall, G. P. McKeown and V. J. Rayward-Smith, ,Cryptanalysis of a Three Rotor Machine Using a Genetic Algorithm,Proceedings of the Seventh International Conference on Genetic Algorithms,1997