Applying sample weighting methods to genetic parallel
programming
S. Cheang, K. Lee, and K. Leung. Proceedings of the 2003 Congress on Evolutionary
Computation CEC2003, page 928--935. Canberra, IEEE Press, (8-12 December 2003)
Abstract
We investigate the sample weighting effect on Genetic
Parallel Programming (GPP). GPP evolves parallel
programs to solve the training samples in a training
set. Usually, the samples are captured directly from a
real-world system. The distribution of samples in a
training set can be extremely biased. Standard GPP
assigns equal weights to all samples. It slows down
evolution because crowded regions of samples dominate
the fitness evaluation causing premature convergence.
This paper presents 4 sample weighting (SW) methods,
i.e. Equal SW, Class-equal SW, Static SW (SSW) and
Dynamic SW (DSW). We evaluate the 4 methods on 7
training sets (3 Boolean functions and 4 UCI medical
data classification databases). Experimental results
show that DSW is superior in performance on all tested
problems. In the 5-input Symmetry Boolean function
experiment, SSW and DSW boost the evolutionary
performance by 465 and 745 times respectively. Due to
the simplicity and effectiveness of SSW and DSW, they
can also be applied to different population-based
evolutionary algorithms.
%0 Conference Paper
%1 Man:2003:Aswmtgpp
%A Cheang, Sin Man
%A Lee, Kin Hong
%A Leung, Kwong Sak
%B Proceedings of the 2003 Congress on Evolutionary
Computation CEC2003
%C Canberra
%D 2003
%E Sarker, Ruhul
%E Reynolds, Robert
%E Abbass, Hussein
%E Tan, Kay Chen
%E McKay, Bob
%E Essam, Daryl
%E Gedeon, Tom
%I IEEE Press
%K algorithms, genetic programming
%P 928--935
%T Applying sample weighting methods to genetic parallel
programming
%X We investigate the sample weighting effect on Genetic
Parallel Programming (GPP). GPP evolves parallel
programs to solve the training samples in a training
set. Usually, the samples are captured directly from a
real-world system. The distribution of samples in a
training set can be extremely biased. Standard GPP
assigns equal weights to all samples. It slows down
evolution because crowded regions of samples dominate
the fitness evaluation causing premature convergence.
This paper presents 4 sample weighting (SW) methods,
i.e. Equal SW, Class-equal SW, Static SW (SSW) and
Dynamic SW (DSW). We evaluate the 4 methods on 7
training sets (3 Boolean functions and 4 UCI medical
data classification databases). Experimental results
show that DSW is superior in performance on all tested
problems. In the 5-input Symmetry Boolean function
experiment, SSW and DSW boost the evolutionary
performance by 465 and 745 times respectively. Due to
the simplicity and effectiveness of SSW and DSW, they
can also be applied to different population-based
evolutionary algorithms.
%@ 0-7803-7804-0
@inproceedings{Man:2003:Aswmtgpp,
abstract = {We investigate the sample weighting effect on Genetic
Parallel Programming (GPP). GPP evolves parallel
programs to solve the training samples in a training
set. Usually, the samples are captured directly from a
real-world system. The distribution of samples in a
training set can be extremely biased. Standard GPP
assigns equal weights to all samples. It slows down
evolution because crowded regions of samples dominate
the fitness evaluation causing premature convergence.
This paper presents 4 sample weighting (SW) methods,
i.e. Equal SW, Class-equal SW, Static SW (SSW) and
Dynamic SW (DSW). We evaluate the 4 methods on 7
training sets (3 Boolean functions and 4 UCI medical
data classification databases). Experimental results
show that DSW is superior in performance on all tested
problems. In the 5-input Symmetry Boolean function
experiment, SSW and DSW boost the evolutionary
performance by 465 and 745 times respectively. Due to
the simplicity and effectiveness of SSW and DSW, they
can also be applied to different population-based
evolutionary algorithms.},
added-at = {2008-06-19T17:35:00.000+0200},
address = {Canberra},
author = {Cheang, Sin Man and Lee, Kin Hong and Leung, Kwong Sak},
biburl = {https://www.bibsonomy.org/bibtex/23f315afe6f7ae12188acde8957e8b54e/brazovayeye},
booktitle = {Proceedings of the 2003 Congress on Evolutionary
Computation CEC2003},
editor = {Sarker, Ruhul and Reynolds, Robert and Abbass, Hussein and Tan, Kay Chen and McKay, Bob and Essam, Daryl and Gedeon, Tom},
interhash = {c584b696d7ad7149c57cf5c2d166cecb},
intrahash = {3f315afe6f7ae12188acde8957e8b54e},
isbn = {0-7803-7804-0},
keywords = {algorithms, genetic programming},
month = {8-12 December},
notes = {CEC 2003 - A joint meeting of the IEEE, the IEAust,
the EPS, and the IEE.},
organisation = {IEEE Neural Network Council (NNC), Engineers Australia
(IEAust), Evolutionary Programming Society (EPS),
Institution of Electrical Engineers (IEE)},
pages = {928--935},
publisher = {IEEE Press},
publisher_address = {445 Hoes Lane, P.O. Box 1331, Piscataway, NJ
08855-1331, USA},
timestamp = {2008-06-19T17:37:36.000+0200},
title = {Applying sample weighting methods to genetic parallel
programming},
year = 2003
}