[ Bristol CS | Index | ML group | Tim Kovacs ] [ Publications ]

Publications by Tim Kovacs Prior to 2006

A more up to date list can be found here.
Where no author is listed, I am the sole author.

Books

Strength or Accuracy? Credit Assignment in Learning Classifier Systems. Springer-Verlag, 2003. ISBN: 1-85233-770-2. (This is a version of my thesis.)
Foundations of Learning Classifier Systems. Bull, L. and Kovacs, T. (editors). The Studies in Fuzziness and Soft Computing series, vol. 183. Springer, 2005. ISBN: 3-540-25073-5.

Journal Articles

Noise, cost and speed-accuracy trade-offs: decision-making in a decentralized system. James A.R. Marshall, Anna Dornhaus, Nigel R. Franks and Tim Kovacs. Journal of The Royal Society Interface, 2005.
Toward a Theory of Generalization and Learning in XCS. Martin Butz, Tim Kovacs, Pier Luca Lanzi and Stewart W. Wilson. IEEE Transactions on Evolutionary Computation 8(1), pages 28-46, 2004.
Rule Fitness and Pathology in Learning Classifier Systems. Evolutionary Computation 12(1), pages 99-135, 2004. ISSN: 1063-6560.
What should a classifier system learn and how should we measure it? The Journal of Soft Computing 6(3-4), pages 171-182. Larry Bull, Pier Luca Lanzi, and Wolfgang Stolzmann, guest editors, 2002.
Learning Classifier Systems Resources. The Journal of Soft Computing 6(3-4), pages 240-243. Larry Bull, Pier Luca Lanzi, and Wolfgang Stolzmann, guest editors, 2002.

Book Chapters

Bibliography of Real-World Classifier Systems Applications. In L. Bull, editor, Applications of Learning Classifier Systems, Studies in Fuzziness and Soft Computing, Vol. 150, pages 300-305. Springer, 2004.
XCS's Strength-Based Twin. Part I. In Lanzi, Stolzmann and Wilson, editors, Learning Classifier Systems, volume LNCS 2661, pages 61--80. Springer, January 2003.
XCS's Strength-Based Twin. Part II. In Lanzi, Stolzmann and Wilson, editors, Learning Classifier Systems, volume LNCS 2661, pages 81--98. Springer, January 2003.
The 2003 Learning Classifier Systems Bibliography. Tim Kovacs and Pier Luca Lanzi. In: Learning Classifier Systems, volume LNCS 2661, pages 187--230. Springer, January 2003.
Two Views of Classifier Systems. In Lanzi, Stolzmann and Wilson, editors, Advances in Learning Classifier Systems, pages 74-87. LNAI 2321, Springer, 2002.
What Makes a Problem Hard for XCS? Tim Kovacs and Manfred Kerber. In Lanzi, Stolzmann, and Wilson (Eds), Advances in Learning Classifier Systems, Lecture Notes in Artificial Intelligence 1996, pp. 80-99. Springer-Verlag, 2001.
abstract/bibtex entry, gzipped postscript
A Bigger Learning Classifier Systems Bibliography. Tim Kovacs and Pier Luca Lanzi. In Lanzi, Stolzmann, and Wilson (Eds), Advances in Learning Classifier Systems, Lecture Notes in Artificial Intelligence 1996, pp. 213-249. Springer-Verlag, 2001.
(Click here for the most up-to-date version in a variety of formats.)
Towards a Theory of Strong Overgeneral Classifiers. In Worthy Martin and William Spears, editors, Foundations of Genetic Algorithms 6, pp. 165-184. Morgan Kaufmann, 2001.
abstact/bibtex entry, gzipped postscript
Strength or Accuracy? Fitness Calculation in Learning Classifier Systems. In Lanzi, Stolzmann, and Wilson (Eds), Learning Classifier Systems. From Foundations to Applications. Lecture Notes in Artificial Intelligence 1813, pp. 143-160. Springer-Verlag, 2000.
abstract, gzipped postscript
A Learning Classifier Systems Bibliography. Tim Kovacs and Pier Luca Lanzi. In Lanzi, Stolzmann, and Wilson (Eds), Learning Classifier Systems. From Foundations to Applications. Lecture Notes in Artificial Intelligence 1813, pp. 321-347. Springer-Verlag, 2000.
(Click here for the most up-to-date version in a variety of formats.)
XCS Classifier System Reliably Evolves Accurate, Complete, and Minimal Representations for Boolean Functions. In Roy, Chawdhry and Pant (Eds), Soft Computing in Engineering Design and Manufacturing (WSC2), pp. 59-68. Springer-Verlag, 1997.
abstract, HTML, and gzipped postscript. (See also the extended technical report version below.)

Conference and Workshop Papers

Evolution of realistic hybrid auctions. Zengchang Qin and Tim Kovacs. Proceedings of the 2004 UK Workshop on Computational Intelligence (UKCI-04), pages 43-50. Mark Withall and Chris Hinde Editors.
High classification accuracy does not imply effective genetic search. Tim Kovacs and Manfred Kerber. In K. Deb et al., GECCO-2004 Proceedings of the 2004 Genetic and Evolutionary Computation Conference, pp. 785-796. Lecture Notes in Computer Science, Vol. 3102, Springer.
A heuristic hill climbing algorithm for mastermind. Alexandre Temporel and Tim Kovacs. Proceedings of the 2003 UK Workshop on Computational Intelligence (UKCI-03), pages 189-196. Jonathan M. Rossiter and Trevor P. Martin Editors. ISBN: 0862925371.
gzipped pdf, powerpoint slides
Simulating the Evolution of Ant Behaviour in Evaluating Nest Sites. James A. R. Marshall, Tim Kovacs, Anna R. Dornhaus, and Nigel R. Franks. In W. Banzhaf, T. Christaller, P. Dittrich, J. T. Kim & J. Ziegler (eds.) Advances in Artificial Life - Proceedings of the 7th European Conference on Artificial Life (ECAL), Lecture Notes in Artificial Intelligence, Vol. 2801, Springer Verlag, Berlin, Heidelberg, 2003, pp. 643-650. © Springer-Verlag. Also in Proceedings of the 2003 UK Workshop on Computational Intelligence (UKCI-03), pages 254-258. Jonathan M. Rossiter and Trevor P. Martin Editors. ISBN: 0862925371.
gzipped pdf, powerpoint slides, LNAI Vol. 2801
Performance and Population State Metrics for Rule-based Learning Systems. Proceedings of the 2002 Congress on Evolutionary Computation (CEC), pages 1781-1786, IEEE Press, 2002.
How XCS Evolves Accurate Classifiers. Martin V. Butz, Tim Kovacs, Pier Luca Lanzi, and Stewart W. Wilson. In Lee Spector et al., editors, proceedings of the Genetic and Evolutionary Computation Conference (GECCO-2001), page 927-934. Morgan Kaufmann. Also tech. Report 2001008, University of Illinois at Urbana-Champaign, 2001.
What should a classifier system learn? In the proceedings of the 2001 Congress on Evolutionary Computation (CEC), volume 2, pages 775-782, IEEE Press, 2001.
What Makes a Problem Hard for XCS? Tim Kovacs and Manfred Kerber. Presented at IWLCS-2000 and as a 2 page extended abstract in the proceedings of PPSN VI.
Slides in gzipped postscript.
Some Dimensions of Problem Complexity for XCS. Tim Kovacs and Manfred Kerber. In the proceedings of the Graduate Student Workshop at GECCO-2000.
abstract.
Strength or Accuracy? A Comparison of Two Approaches to Fitness Calculation in Learning Classifier Systems. Presented at the 2nd International Workshop on Learning Classifier Systems (IWLCS-99), and appeared in the Proceedings of the 1999 Genetic and Evolutionary Computation Conference Workshop Program, Annie S. Wu (Ed.), pp. 258-265, 1999.
(The extended book chapter version is much improved over this one.)
Deletion Schemes for Classifier Systems. In W. Banzhaf, J. Daida, A. E. Eiben, M. H. Garzon, V. Honavar, M. Jakiela, and R. E. Smith, editors, GECCO-99: Proceedings of the Genetic and Evolutionary Computation Conference, pp. 329-336. Morgan Kaufmann, San Francisco (CA), 1999.
abstract (Technical report with minor corrections available below.)

Technical Reports

(Only reports which are not identical to papers listed elsewhere on this page are shown.)

A Proposal for Population-Based Reinforcement Learning. Tim Kovacs and Stuart I. Reynolds. Technical Report CSTR-03-001, Department of Computer Science, University of Bristol, January 2003.
Trends in Learning Classifier Systems Publication. Technical Report CSRP-00-21, School of Computer Science, University of Birmingham, 2000.
abstract/bibtex entry gzipped postscript
Deletion Schemes for Classifier Systems. A Technical Report. CSRP-99-08, School of Computer Science, University of Birmingham.
abstract/bibtex entry, gzipped postscript
XCS Classifier System Reliably Evolves Accurate, Complete, and Minimal Representations for Boolean Functions. Technical Report Version. CSRP-97-19, School of Computer Science, University of Birmingham.
abstract/bibtex entry and gzipped postscript copy of the tech report version (and an html copy of the conference paper version).
For some reason I forgot to include a figure showing condensation and subset extraction working on the 6 bit even parity problem in the technical report, but here it is now.

Theses

A Comparison of Strength and Accuracy-based Fitness in Learning Classifier Systems.
Tim Kovacs, 2002. PhD Thesis. School of Computer Science. University of Birmingham. Birmingham, U.K. Supervisor: Manfred Kerber.
[Thesis page]
Evolving Optimal Populations with XCS Classifier Systems.
Tim Kovacs, 1996. MSc Advanced Computer Science Thesis. School of Computer Science. University of Birmingham. Birmingham, U.K. Supervisors: Aaron Sloman and Ian Wright.
(This also appeared as technical report CSRP-96-17 / CSR-96-17 of the school of computer science, U. of Birmingham.)
abstract/bibtex entry, gzipped postscript copy and errata in gzipped postscript
A Comparison of Human and Computer Scoring of Periodic Limb Movements in Sleep.
Tim Kovacs, 1995. B.A. Honours Psychology Thesis. Carleton University, Ottawa, Ontario, Canada. Supervisor: Rick Bonato.

Miscellaneous

Cognition and Affect Project Summary (1996).
I've written a summary of the Cognition and Affect Project led by Aaron Sloman at Birmingham.
html and gzipped postscript