ABSTRACT
This paper investigates the effect on building blocks during evolution of two online program simplification methods in genetic programming. The two simplification methods considered are algebraic simplification and numerical simplification. The building blocks considered are of a more general form (two and three level subtrees) than numeric constants only. Unlike most of the existing work which often uses simple symbolic regression tasks, this work considers classification tasks as examples. We develop a new method for encoding possible building blocks for the analysis. The results show that the two online program simplification methods can generate new diverse building blocks during evolution although they also destroy existing ones and that many of the existing building blocks are retained during evolution. Compared with the canonical genetic programming method, the two simplification methods can generate much smaller programs, use much shorter evolutionary training time and achieve comparable effectiveness performance.
- ]]W. Banzhaf, P. Nordin, R. E. Keller, and F. D. Francone. Genetic Programming: An Introduction on the Automatic Evolution of computer programs and its Applications. Morgan Kaufmann Publishers. 1998. Google ScholarDigital Library
- ]]M. Brameier and W. Banzhaf. A comparison of linear genetic programming and neural networks in medical data mining. IEEE Transactions on Evolutionary Computation, 5(1):17--26, feb 2001. Google ScholarDigital Library
- ]]A. Ekart. Shorter fitness preserving genetic programs. In C. Fonlupt, et al. editors, Proccedings of 4th European Conference on Artificial Evolution, volume 1829 of LNCS, pages 73--83, 2000. Google ScholarDigital Library
- ]]M. Forina, and R. Leardi, C. Armanino, and S. Lanteri. Parvus: an extendable package of programs for data exploration, classification and correlation. Elsevier Scientific, 1988. Amsterdam, The Netherlands.Google Scholar
- ]]S. Gustafson, A. Ekart, E. Burke, and G. Kendall. Problem difficulty and code growth in genetic programming. Genetic Programming and Evolvable Machines, 5(3):271--290, Sept. 2004. Google ScholarDigital Library
- ]]D. Hooper and N. S. Flann. Improving the accuracy and robustness of genetic programming through expression simplification. In J. R. Koza, et al. editors, Genetic Programming 1996: Proceedings of the First Annual Conference, page 428, 1996. Google ScholarDigital Library
- ]]D. Kinzett, M. Zhang, and M. Johnston. Using numerical simplification to control bloat in genetic programming. In Proceedings of the 7th International Conference Simulated Evolution and Learning (SEAL2008), pages 493--502, 2008. Google ScholarDigital Library
- ]]J. R. Koza. Genetic Programming: On the Programming of Computers by Means of Natural Selection. MIT Press, Cambridge, MA, USA, 1992. Google ScholarDigital Library
- ]]J. R. Koza, M. A. Keane, M. J. Streeter, W. Mydlowec, J. Yu, and G. Lanza. Genetic Programming IV: Routine Human-Competitive Machine Intelligence. Kluwer Academic Publishers, 2003. Google ScholarDigital Library
- ]]W. B. Langdon. Quadratic bloat in genetic programming. In D. Whitley, et al., editors, Proceedings of the Genetic and Evolutionary Computation Conference (GECCO-2000), pages 451--458, 2000. Morgan Kaufmann.Google Scholar
- ]]W. B. Langdon and R. Poli. Fitness causes bloat. In P. K. Chawdhry, et al., editors, Soft Computing in Engineering Design and Manufacturing, pages 13--22. Springer-Verlag London, 23-27 June 1997.Google Scholar
- ]]S. Luke and L. Panait. Lexicographic parsimony pressure. In W. B. Langdon, et al., editors, GECCO 2002: Proceedings of the Genetic and Evolutionary Computation Conference, pages 829--836, 2002.Google Scholar
- ]]S. Luke and L. Panait. A comparison of bloat control methods for genetic programming. Evolutionary Computation, 14(3):309--344, Fall 2006. Google ScholarDigital Library
- ]]D. Marshall. The discrete cosine transform. http://www.cs.cf.ac.uk/Dave/Multimedia/node231.htm, 2001.Google Scholar
- ]]P. Nordin and W. Banzhaf. Complexity compression and evolution. In L. Eshelman, editor, Genetic Algorithms: Proceedings of the Sixth International Conference (ICGA95), pages 310--317, 1995. Morgan Kaufmann. Google ScholarDigital Library
- ]]D. Parrott, X. Li, and V. Ciesielski. Multi-objective techniques in genetic programming for evolving classifiers. In D. Corne, et al., editors, Proceedings of the 2005 IEEE Congress on Evolutionary Computation, volume 2, pages 1141--1148, 2005.Google ScholarCross Ref
- ]]R. Poli, W. B. Langdon, and N. F. McPhee. A field guide to genetic programming. Published via http://lulu.com and freely available at http://www.gp-field-guide.org.uk, 2008. Google ScholarDigital Library
- ]]S. Silva and E. Costa. Dynamic limits for bloat control in genetic programming and a review of past and current bloat theories. Genetic Programming and Evolvable Machines. 2009. Online First. Google ScholarDigital Library
- ]]J. F. Smith, III. Genetic program based data mining for fuzzy decision trees. In Z. R. Yang, et al. editors, Intelligent Data Engineering and Automated Learning - IDEAL 2004, 5th International Conference, Proceedings, pages 464--470, 2004. Springer.Google Scholar
- ]]T. Soule and J. A. Foster. Effects of code growth and parsimony pressure on populations in genetic programming. Evolutionary Computation, 6(4):293--309, Winter 1998. Google ScholarDigital Library
- ]]M. J. Streeter. The root causes of code growth in genetic programming. In C. Ryan, et al. editors, Genetic Programming, Proceedings of Euro GP'2003, pages 443--454, 2003. Springer-Verlag. Google ScholarDigital Library
- ]]I. Witten and E. Frank. Data Mining: Practical machine learning tools and techniques. Morgan Kaufmann, San Francisco, 2 edition, 2005. Google ScholarDigital Library
- ]]P. Wong and M. Zhang. Algebraic simplification of GP programs during evolution. In M. Keijzer, et al., editors, GECCO 2006: Proceedings of the 8th annual conference on Genetic and evolutionary computation, volume 1, pages 927--934, 2006. ACM Press. Google ScholarDigital Library
- ]]P. Wong and M. Zhang. Effects of program simplification on simple building blocks in genetic programming. In IEEE Congress on Evolutionary Computation, pages 1570--1577, 2007.Google Scholar
- ]]B.-T. Zhang and H. M¨ uhlenbein. Balancing accuracy and parsimony in genetic programming. Evolutionary Computation, 3(1):17--38, 1995. Google ScholarDigital Library
- ]]M. Zhang, Y. Zhang, and W. D. Smart. Program simplification in genetic programming for object classification. In R. Khosla, et al., editors, Proceedings of the 9th International Conference Knowledge-Based Intelligent Information and Engineering Systems, Part III, pages 988--996, 2005. Springer. Google ScholarDigital Library
Index Terms
- How online simplification affects building blocks in genetic programming
Recommendations
Numerical-node building block analysis of genetic programming with simplification
GECCO '07: Proceedings of the 9th annual conference on Genetic and evolutionary computationThis paper investigates the effects on building blocks of using online simplification in a GP system. Numerical nodes are tracked through individual runs to observe their behaviour. Results show that simplification disrupts building blocks early on, but ...
Building Blocks Identification Based on Subtree Sample Counts for Genetic Programming
APCASE '15: Proceedings of the 2015 Asia-Pacific Conference on Computer Aided System EngineeringOften, the performance of genetic programming (GP) is explained in terms of building blocks--high-quality solution elements that get gradually assembled into larger and more complex patterns by the evolutionary process. However, the weak theoretical ...
Estimating the distribution and propagation of genetic programming building blocks through tree compression
GECCO '09: Proceedings of the 11th Annual conference on Genetic and evolutionary computationShin et al [19] and McKay et al [15] previously applied tree compression and semantics-based simplification to study the distribution of building blocks in evolving Genetic Programming populations. However their method could only give static estimates ...
Comments