%0 Conference Paper %1 ondas:2005:WSC %A Ondas, Radovan %A Pelikan, Martin %A Sastry, Kumara %B WSC10: 10th Online World Conference on Soft Computing in Industrial Applications %C On the World Wide Web %D 2005 %E Knowles, Joshua %K PIPE, algorithms, genetic order problem problem, programming, scalability, scaling, trap %P 363--372 %T Genetic Programming, Probabilistic Incremental Program Evolution, and Scalability %U http://isxp1010c.sims.cranfield.ac.uk/Papers/paper122.pdf %X We discuss scalability of standard genetic programming (GP) and the probabilistic incremental program evolution (PIPE). To investigate the need for both effective mixing and linkage learning, two test problems are considered: ORDER problem, which is rather easy for any recombination-based GP, and TRAP or the deceptive trap problem, which requires the algorithm to learn interactions among subsets of terminals. The scalability results show that both GP and PIPE scale up polynomially with problem size on the simple ORDER problem, but they both scale up exponentially on the deceptive problem. This indicates that while standard recombination is sufficient when no interactions need to be considered, for some problems linkage learning is necessary. These results are in agreement with the lessons learned in the domain of binary-string genetic algorithms (GAs). Furthermore, the paper investigates the effects of introducing unnecessary and irrelevant primitives on the performance of GP and PIPE. %@ 3-540-29123-7

@inproceedings{ondas:2005:WSC, abstract = {We discuss scalability of standard genetic programming (GP) and the probabilistic incremental program evolution (PIPE). To investigate the need for both effective mixing and linkage learning, two test problems are considered: ORDER problem, which is rather easy for any recombination-based GP, and TRAP or the deceptive trap problem, which requires the algorithm to learn interactions among subsets of terminals. The scalability results show that both GP and PIPE scale up polynomially with problem size on the simple ORDER problem, but they both scale up exponentially on the deceptive problem. This indicates that while standard recombination is sufficient when no interactions need to be considered, for some problems linkage learning is necessary. These results are in agreement with the lessons learned in the domain of binary-string genetic algorithms (GAs). Furthermore, the paper investigates the effects of introducing unnecessary and irrelevant primitives on the performance of GP and PIPE.}, added-at = {2008-06-19T17:46:40.000+0200}, address = {On the World Wide Web}, author = {Ondas, Radovan and Pelikan, Martin and Sastry, Kumara}, biburl = {https://www.bibsonomy.org/bibtex/22d9f8f78cdd0b609b4fc59b372eef1a4/brazovayeye}, booktitle = {WSC10: 10th Online World Conference on Soft Computing in Industrial Applications}, editor = {Knowles, Joshua}, email = {ondasr@umsl.edu pelikan@cs.umsl.edu ksastry@uiuc.edu}, interhash = {499afba369febbcc812f8f60da3e0590}, intrahash = {2d9f8f78cdd0b609b4fc59b372eef1a4}, isbn = {3-540-29123-7}, keywords = {PIPE, algorithms, genetic order problem problem, programming, scalability, scaling, trap}, month = {19 September - 7 October}, notes = {http://www.cranfield.ac.uk/wsc10/ broken Cf \cite{1068310}. Experimental, ORDER, TRAP, JOIN, NEG_JOIN, JUNK. Linkage learning and EDAs should be applied to GP. http://isxp1010c.sims.cranfield.ac.uk/Presentations/presentation122.pdf broken slides (1.3Mbyte)}, organisation = {World Federation of Soft Computing (WFSC), European Neural Network Society (ENNS), North American Fuzzy Information Processing Society (NAFIPS), European Society for Fuzzy Logic and Technology (EUSFLAT), and International Fuzzy Systems Association (IFSA)}, pages = {363--372}, size = {10 pages}, timestamp = {2008-06-19T17:48:55.000+0200}, title = {Genetic Programming, Probabilistic Incremental Program Evolution, and Scalability}, url = {http://isxp1010c.sims.cranfield.ac.uk/Papers/paper122.pdf}, year = 2005 }