Sequential model building in symbolic regression

0512085 - ÚI 2020 RIV DE eng C - Konferenční příspěvek (zahraniční konf.)
Žegklitz, Jan - Pošík, M.
Sequential model building in symbolic regression.
ITAT 2019: Information Technologies – Applications and Theory. Aachen: Technical University & CreateSpace Independent Publishing, 2019 - (Barančíková, P.; Holeňa, M.; Horváth, T.; Pleva, M.; Rosa, R.), s. 51-57. CEUR Workshop Proceeding, 2473. ISSN 1613-0073.
[ITAT 2019: Conference Information Technologies - Applications and Theory /19./. Donovaly (SK), 20.09.2019-24.09.2019]
Grant CEP: GA ČR GA17-01251S
Grant ostatní: GA MŠk(CZ) LM2015042
Institucionální podpora: RVO:67985807
Klíčová slova: genetic programming * symbolic regression * boosting * sequential learning
Obor OECD: Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
http://ceur-ws.org/Vol-2473/paper5.pdf

Symbolic Regression is a supervised learning technique for regression based on Genetic Programming. A popular algorithm is the Multi-Gene Genetic Programming which builds models as a linear combination of a number of components which are all built together. However, in recent years a different approach emerged, represented by the Sequential Symbolic Regression algorithm, which builds the model sequentially, one component at a time, and the components are combined using a method based on geometric semantic crossover. In this article we show that the SSR algorithm effectively produces linear combination of components and we introduce another sequential approach very similar to classical ensemble method of boosting. All algorithms are compared with MGGP as a baseline on a number of real-world datasets. The results show that the sequential approaches are overall worse than MGGP both in terms of accuracy and model size.

Trvalý link: http://hdl.handle.net/11104/0302291