Probability Based Genetic Programming for Multiclass
Object Classification
W. Smart, and M. Zhang. CS-TR-04-7. Computer Science, Victoria University of Wellington, New Zealand, (2004)
Abstract
Instead of using predefined multiple thresholds to
form different regions in the program output space for
different classes, this approach uses probabilities of
different classes, derived from Gaussian distributions,
to construct the fitness function for classification.
Two fitness measures, overlap area and weighted
distribution distance, have been developed. The
approach is examined on three multiclass object
classification problems of increasing difficulty and
compared with a basic GP approach. The results suggest
that the new approach is more effective and more
efficient than the basic GP approach. While the area
measure was a bit more effective than the distance
measure in most cases, the distance measure was more
efficient to learn good program classifiers.
%0 Report
%1 vuw-CS-TR-04-7
%A Smart, Will
%A Zhang, Mengjie
%C New Zealand
%D 2004
%K Gaussian Probability algorithms, area, based classification distance, distribution distribution, genetic multiclass object overlap programming, weighted
%N CS-TR-04-7
%T Probability Based Genetic Programming for Multiclass
Object Classification
%U http://www.mcs.vuw.ac.nz/comp/Publications/CS-TR-04-7.abs.html
%X Instead of using predefined multiple thresholds to
form different regions in the program output space for
different classes, this approach uses probabilities of
different classes, derived from Gaussian distributions,
to construct the fitness function for classification.
Two fitness measures, overlap area and weighted
distribution distance, have been developed. The
approach is examined on three multiclass object
classification problems of increasing difficulty and
compared with a basic GP approach. The results suggest
that the new approach is more effective and more
efficient than the basic GP approach. While the area
measure was a bit more effective than the distance
measure in most cases, the distance measure was more
efficient to learn good program classifiers.
@techreport{vuw-CS-TR-04-7,
abstract = {Instead of using predefined multiple thresholds to
form different regions in the program output space for
different classes, this approach uses probabilities of
different classes, derived from Gaussian distributions,
to construct the fitness function for classification.
Two fitness measures, overlap area and weighted
distribution distance, have been developed. The
approach is examined on three multiclass object
classification problems of increasing difficulty and
compared with a basic GP approach. The results suggest
that the new approach is more effective and more
efficient than the basic GP approach. While the area
measure was a bit more effective than the distance
measure in most cases, the distance measure was more
efficient to learn good program classifiers.},
added-at = {2008-06-19T17:35:00.000+0200},
address = {New Zealand},
author = {Smart, Will and Zhang, Mengjie},
biburl = {https://www.bibsonomy.org/bibtex/27bd5b55c339d76fdb5b5ce8ad9bb40ce/brazovayeye},
institution = {Computer Science, Victoria University of Wellington},
interhash = {187409c4ff03cfcaa7fa84fd904b1c5c},
intrahash = {7bd5b55c339d76fdb5b5ce8ad9bb40ce},
keywords = {Gaussian Probability algorithms, area, based classification distance, distribution distribution, genetic multiclass object overlap programming, weighted},
number = {CS-TR-04-7},
timestamp = {2008-06-19T17:51:48.000+0200},
title = {Probability Based Genetic Programming for Multiclass
Object Classification},
url = {http://www.mcs.vuw.ac.nz/comp/Publications/CS-TR-04-7.abs.html},
year = 2004
}