Preconditions of GPA-ES Algorithm Application to Big Data
- 21 Downloads
Herein presented contribution speaks about preconditions of genetic programming algorithm (GPA) used to work with so-called big data. In the paper, the different ways of decrease of number of (big) training data evaluations as well as parallelization of this process are presented. The paper also discusses using a floating window to provide evolution steps on small data subsets. The presented approach is based on reduction of the number of evaluations to allow fitness function evaluation in acceptable computational time. The original GPA-ES algorithm is described too. This algorithm was developed and tested for many years for application in area of accurate symbolic regression. Now the interest is to apply it to the application domain of big data, especially to data-based modeling and knowledge discovering fields. Then, there are described experiments verifying these hypotheses and the results are discussed. These experiments are based on symbolic regression (discovering of differential equations) describing a training data set representing movement of the Lorenz attractor deterministic chaos system.
KeywordsBig data Evolutionary algorithm Evaluation scheme Evaluation reduction Floating window Data subset
This work was supported by The Ministry of Education, Youth and Sports from the Large Infrastructures for Research, Experimental Development and Innovations project “IT4Innovations National Super-computing Center—LM2015070”.
- 1.Technopedia: Big data. https://www.techopedia.com/definition/27745/big-data (2019). Accessed 11 March 2019
- 2.Nirvanshu, H., Sana, M., Omkar, S.N.: Big data clustering using genetic algorithm on hadoop mapreduce. Inter. J. Sci. Technol. Res. 4(4), (2015). ISSN 2277-8616Google Scholar
- 3.Bader-EL-Den, M., Adda, M.: Automatic programming for optimization problems with big data. https://www.findaphd.com/phds/project/automatic-programming-for-optimization-problems-with-big-data/?p62613 (2019). Accessed 12 March 2019
- 6.Andre, D., Koza, J.R.: Parallel genetic programming: A scalable implementation using the transputer network architecture. In: Angeline, J.P., Kinnear Jr, K.E. (eds) Advances in Genetic Programming 2, The MIT Press, (1996)Google Scholar
- 7.Hugues, J., Pollack, J., B.: Pararllel genetic programming and fine-grained SIMD architecture. In: Siegel, E.V., Koza, J.R. (eds.) Working Notes for the AAAI Symposium on Genetic Programming, pp. 31–37. MIT, Cambridge (1995)Google Scholar
- 9.Vekaria, K., Clack, C.: Haploid genetic programming with dominance. Technical Report RN/97/121, (1997)Google Scholar
- 10.Brandejsky, T.: Small populations in GPA-ES algorithm. In: Matousek, R. (ed.) 19th International Conference on Soft Computing Mendel 2013, pp. 31–36, Brno University of Technology (2013). ISSN 1803-3814. ISBN 978-80-214-4755-4Google Scholar
- 12.Brandejsky, T.: Multi-layered evolutionary system suitable to symbolic model regression. In: Recent Researches in Applied Informatics, Praha, 2011-09-26/2011-09-28, pp. 222–225. WSEAS Press, Athens (2011). ISBN 978-1-61804-034-3Google Scholar