Presentation 1999/5/14
Dynamic Growth Control of Trees in Genetceterasic Prograrmming
Noritaka SHIGEI, Motonori TAYA, Hiromi MIYAJIMA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In traditional genetic programming (GP), the explosive growth of trees often occurs, and it requires an enormous amount of calculation. To conquer this difficulty, we consider fitness functions based on error and complexity terms. Although the fitness functions can restrain growth of tree, the balance of error and complexity terms is important to prevent excessive restraint of tree growth. In this paper, we present a new method dynamically balancing error and complexity terms. Our method is compared with the method already proposed by Zhang et al. The effectiveness of our method is shown by a computer simulation on symbolic regressions.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) genetic programming / growth control of trees / fitness / error / description length / symbolic regression
Paper # NLP99-12
Date of Issue

Conference Information
Committee NLP
Conference Date 1999/5/14(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Nonlinear Problems (NLP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Dynamic Growth Control of Trees in Genetceterasic Prograrmming
Sub Title (in English)
Keyword(1) genetic programming
Keyword(2) growth control of trees
Keyword(3) fitness
Keyword(4) error
Keyword(5) description length
Keyword(6) symbolic regression
1st Author's Name Noritaka SHIGEI
1st Author's Affiliation Shimane University()
2nd Author's Name Motonori TAYA
2nd Author's Affiliation Nara Institute of Science and Technology
3rd Author's Name Hiromi MIYAJIMA
3rd Author's Affiliation Kagoshima University
Date 1999/5/14
Paper # NLP99-12
Volume (vol) vol.99
Number (no) 42
Page pp.pp.-
#Pages 7
Date of Issue