Employing Nominal Attributes in Classification Using Genetic Programming

Created by W.Langdon from gp-bibliography.bib Revision:1.3973

@InProceedings{loveard:2002:SEAL,
  author =       "Thomas Loveard and Vic Ciesielski",
  title =        "Employing Nominal Attributes in Classification Using
                 Genetic Programming",
  booktitle =    "Proceedings of the 4th Asia-Pacific Conference on
                 Simulated Evolution And Learning (SEAL'02)",
  year =         "2002",
  editor =       "Lipo Wang and Kay Chen Tan and Takeshi Furuhashi and 
                 Jong-Hwan Kim and Xin Yao",
  pages =        "487--491",
  address =      "Orchid Country Club, Singapore",
  month =        "18-22 " # nov,
  publisher =    "IEEE",
  keywords =     "genetic algorithms, genetic programming",
  ISBN =         "981-04-7522-5",
  URL =          "http://goanna.cs.rmit.edu.au/~vc/papers/seal02-loveard.pdf",
  size =         "5 pages",
  abstract =     "In this paper methods for performing classification
                 using Genetic Programming (GP) on datasets with nominal
                 attributes are developed and evaluated. The two methods
                 developed included the splitting of GP program
                 execution based upon the value of a nominal attribute
                 (execution branching), and the conversion of a nominal
                 attribute to a continuous or binary attribute (numeric
                 conversion). These two methods of using nominal
                 attributes are tested against six datasets containing
                 either nominal and continuous attributes or nominal
                 only attributes.

                 Results show that the use of the methods developed in
                 this paper allow classifiers trained with GP to perform
                 accurate classification of datasets containing nominal
                 attributes. When compared to other well-known methods
                 of classification the GP method is capable of
                 classifying one of six datasets more accurately than
                 any of the conventional methods tested, and accuracy
                 close to the best achieved method on 3 other
                 datasets.",
  notes =        "SEAL 2002 see
                 http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.200.6410&rep=rep1&type=pdf",
}

Genetic Programming entries for Thomas Loveard Victor Ciesielski

Citations