Abstract
Function approximation is an important technique used in many different domains, including numerical mathematics, engineering, and neuroscience. The XCSF classifier system is able to approximate complex multi-dimensional function surfaces using a patchwork of simpler functions. Typically, locally linear functions are used due to the tradeoff between expressiveness and interpretability. This work discusses XCSF’s current capabilities, but also points out current challenges that can hinder learning success. A theoretical discussion on when XCSF works is intended to improve the comprehensibility of the system. Current advances with respect to scalability theory show that the system constitutes a very effective machine learning technique. Furthermore, the paper points-out how to tune relevant XCSF parameters in actual applications and how to choose appropriate condition and prediction structures. Finally, a brief comparison to the Locally Weighted Projection Regression (LWPR) algorithm highlights positive as well as negative aspects of both methods.
Keywords
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Holland, J.H.: Adaptation in natural and artificial systems: An introductory analysis with applications to biology, control, and artificial intelligence. The MIT Press, Cambridge (1992)
Wilson, S.W.: Classifier fitness based on accuracy. Evolutionary Computation 3(2), 149–175 (1995)
Butz, M.V., Goldberg, D.E., Lanzi, P.L.: Gradient descent methods in learning classifier systems: Improving XCS performance in multistep problems. Technical report, Illinois Genetic Algorithms Laboratory (2003)
Bernadó-Mansilla, E., Garrell-Guiu, J.M.: Accuracy-based learning classifier systems: Models, analysis, and applications to classification tasks. Evolutionary Computation 11, 209–238 (2003)
Butz, M.V.: Rule-Based Evolutionary Online Learning Systems: A Principal Approach to LCS Analysis and Design. Springer, Heidelberg (2006)
Butz, M.V., Herbort, O.: Context-dependent predictions and cognitive arm control with XCSF. In: GECCO 2008: Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation, pp. 1357–1364. ACM, New York (2008)
Stalph, P.O., Butz, M.V., Pedersen, G.K.M.: Controlling a four degree of freedom arm in 3D using the XCSF learning classifier system. In: Mertsching, B., Hund, M., Aziz, Z. (eds.) KI 2009. LNCS, vol. 5803, pp. 193–200. Springer, Heidelberg (2009)
Wilson, S.W.: Classifiers that approximate functions. Natural Computing 1, 211–234 (2002)
Stalph, P.O., Llorà, X., Goldberg, D.E., Butz, M.V.: Resource Management and Scalability of the XCSF Learning Classifier System. Theoretical Computer Science (in press), http://dx.doi.org/10.1016/j.tcs.2010.07.007
Butz, M.V., Kovacs, T., Lanzi, P.L., Wilson, S.W.: How XCS evolves accurate classifiers. In: Proceedings of the Genetic and Evolutionary Computation Conference (GECCO 2001), pp. 927–934 (2001)
Wright, A.H.: Genetic algorithms for real parameter optimization. In: Foundations of Genetic Algorithms, pp. 205–218. Morgan Kaufmann, San Francisco (1991)
Goldberg, D.E.: Real-coded genetic algorithms, virtual alphabets, and blocking. Complex Systems 5, 139–167 (1991)
Radcliffe, N.J.: Equivalence class analysis of genetic algorithms. Complex Systems 5, 183–205 (1991)
Mühlenbein, H., Schlierkamp-Voosen, D.: Predictive models for the breeder genetic algorithm – I. continuous parameter optimization. Evolutionary Computation 1, 25–49 (1993)
Beyer, H.G., Schwefel, H.P.: Evolution strategies - a comprehensive introduction. Natural Computing 1(1), 3–52 (2002)
Bosman, P.A.N., Thierens, D.: Numerical optimization with real-valued estimation-of-distribution algorithms. In: Scalable Optimization via Probabilistic Modeling. SCI, vol. 33, pp. 91–120. Springer, Heidelberg (2006)
Stalph, P.O., Butz, M.V.: How Fitness Estimates Interact with Reproduction Rates: Towards Variable Offspring Set Sizes in XCSF. In: Bacardit, J. (ed.) IWLCS 2008/2009. LNCS (LNAI), vol. 6471, pp. 47–56. Springer, Heidelberg (2010)
Orriols-Puig, A., Bernadó-Mansilla, E.: Bounding XCS’s parameters for unbalanced datasets. In: GECCO 2006: Proceedings of the 8th Annual Conference on Genetic and Evolutionary Computation, pp. 1561–1568. ACM, New York (2006)
Kovacs, T., Kerber, M.: What makes a problem hard for XCS? In: Lanzi, P.L., Stolzmann, W., Wilson, S.W. (eds.) IWLCS 2000. LNCS (LNAI), vol. 1996, pp. 251–258. Springer, Heidelberg (2001)
Wilson, S.W.: Get real! XCS with continuous-valued inputs. In: Lanzi, P.L., Stolzmann, W., Wilson, S.W. (eds.) IWLCS 1999. LNCS (LNAI), vol. 1813, pp. 209–219. Springer, Heidelberg (2000)
Wilson, S.W.: Generalization in the XCS classifier system. In: Genetic Programming 1998: Proceedings of the Third Annual Conference, pp. 665–674 (1998)
Stone, C., Bull, L.: For real! XCS with continuous-valued inputs. Evolutionary Computation 11(3), 299–336 (2003)
Butz, M.V., Lanzi, P.L., Wilson, S.W.: Function approximation with XCS: Hyperellipsoidal conditions, recursive least squares, and compaction. IEEE Transactions on Evolutionary Computation 12, 355–376 (2008)
Wilson, S.W.: Classifier conditions using gene expression programming. In: Bacardit, J., Bernadó-Mansilla, E., Butz, M.V., Kovacs, T., Llorà, X., Takadama, K. (eds.) IWLCS 2006 and IWLCS 2007. LNCS (LNAI), vol. 4998, pp. 206–217. Springer, Heidelberg (2008)
Lanzi, P.L., Loiacono, D., Wilson, S.W., Goldberg, D.E.: Extending XCSF beyond linear approximation. In: GECCO 2005: Proceedings of the 2005 Conference on Genetic and Evolutionary Computation, pp. 1827–1834 (2005)
Vijayakumar, S., Schaal, S.: Locally weighted projection regression: An O(n) algorithm for incremental real time learning in high dimensional space. In: ICML 2000: Proceedings of the Seventeenth International Conference on Machine Learning, pp. 1079–1086 (2000)
Vijayakumar, S., D’Souza, A., Schaal, S.: Incremental online learning in high dimensions. Neural Computation 17(12), 2602–2634 (2005)
Stalph, P.O., Rubinsztajn, J., Sigaud, O., Butz, M.V.: A comparative study: Function approximation with LWPR and XCSF. In: GECCO 2010: Proceedings of the 12th Annual Conference on Genetic and Evolutionary Computation (in press, 2010)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2010 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Stalph, P.O., Butz, M.V. (2010). Current XCSF Capabilities and Challenges. In: Bacardit, J., Browne, W., Drugowitsch, J., Bernadó-Mansilla, E., Butz, M.V. (eds) Learning Classifier Systems. IWLCS IWLCS 2009 2008. Lecture Notes in Computer Science(), vol 6471. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-17508-4_5
Download citation
DOI: https://doi.org/10.1007/978-3-642-17508-4_5
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-17507-7
Online ISBN: 978-3-642-17508-4
eBook Packages: Computer ScienceComputer Science (R0)