Abstract
State of the art equation discovery systems start the discovery process from scratch, rather than from an initial hypothesis in the space of equations. On the other hand, theory revision systems start from a given theory as an initial hypothesis and use new examples to improve its quality. Two quality criteria are usually used in theory revision systems. The first is the accuracy of the theory on new examples and the second is the minimality of change of the original theory. In this paper, we formulate the problem of theory revision in the context of equation discovery. Moreover, we propose a theory revision method suitable for use withth e equation discovery system Lagramge. The accuracy of the revised theory and the minimality of theory change are considered. The use of the method is illustrated on the problem of improving an existing equation based model of the net production of carbon in the Earth ecosystem. Experiments show that small changes in the model parameters and structure considerably improve the accuracy of the model.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
P. Langley, H. A. Simon, G. L. Bradshaw, and J. M. Żythow. Scientific Discovery.MIT Press, Cambridge, MA, 1987.
N. Lavrac and Sašo Džeroski. Inductive Logic Programming: Techniques and Applications.Ellis Horwood, Chichester, 1994. Freely available athttp://www-ai.ijs.si/SasoDzeroski/ILPBook/.
D. Ourston and R. J. Mooney. Theory refinement combining analytical and empiricalmethods. Artificial Intelligence, 66:273–309, 1994.
C. S. Potter and S. A. Klooster. Interannual variability in soil trace gas (CO2,N2O, NO) uxes and analysis of controllers on regional to global scales. GlobalBiogeochemical Cycles, 12:621–635, 1998.
K. Saito, P. Langley, and T. Grenager.The computational revision of quantitativescientific models. 2001. Submitted to Discovery Science conference.
L. Todorovski and S. Džeroski. Declarative bias in equation discovery. In Proceedingsofthe Fourteenth International Conference on Machine Learning, pages376–384, Nashville, MA, 1997. Morgan Kaufmann.
T. Washio and H. Motoda. Discovering admissible models of complex systemsbased on scale-types and identity constraints. In Proceedings of the Fifteenth InternationalJoint Conference on Artificial Intelligence, volume 2, pages 810–817,Nogoya, Japan, 1997. Morgan Kaufmann.
P. A. Whigham and F. Recknagel. Predicting chlorophyll-a in freshwater lakesby hybridising process-based models and genetic algorithms. In Book ofA bstractsofthe Second International Conference on Applications ofMachine Learning toEcological Modeling. Adelaide University, 2000.
S. Wrobel. First order theory refinement. InL.Y De Raedt, editor, Advances inInductive Logic Programming, pages 14–33. IOS Press, 1996.
R. Zembowicz and J. M. Żytkow. Discovery of equations: Experimental evaluationof convergence. In Proceedings ofthe Tenth National Conference on ArtificialIntelligence, pages 70–75, San Jose, CA, 1992. Morgan Kaufmann.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2001 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Todorovski, L., Džeroski, S. (2001). Theory Revision in Equation Discovery. In: Jantke, K.P., Shinohara, A. (eds) Discovery Science. DS 2001. Lecture Notes in Computer Science(), vol 2226. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-45650-3_33
Download citation
DOI: https://doi.org/10.1007/3-540-45650-3_33
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-42956-2
Online ISBN: 978-3-540-45650-6
eBook Packages: Springer Book Archive