- author = "Thomas Loveard and Victor Ciesielski",
- title = "Representing Classification Problems in Genetic Programming",
- booktitle = "Proceedings of the Congress on Evolutionary Computation",
- year = "2001",
- volume = "2",
- pages = "1070--1077",
- address = "COEX, World Trade Center, 159 Samseong-dong, Gangnam-gu, Seoul, Korea",
- publisher_address = "445 Hoes Lane, P.O. Box 1331, Piscataway, NJ 08855-1331, USA",
- month = "27-30 " # may,
- organisation = "IEEE Neural Network Council (NNC), Evolutionary Programming Society (EPS), Institution of Electrical Engineers (IEE)",
- publisher = "IEEE Press",
- email = "toml@cs.rmit.edu.au",
- keywords = "genetic algorithms, genetic programming, Classification",
- ISBN = "0-7803-6657-3",
-
URL = "
http://goanna.cs.rmit.edu.au/~toml/cec2001.ps",
-
DOI = "
doi:10.1109/CEC.2001.934310",
- abstract = "In this paper five alternative methods are proposed to perform multi-class classification tasks using genetic programming. These methods are: Binary decomposition, in which the problem is decomposed into a set of binary problems and standard genetic programming methods are applied; Static range selection, where the set of real values returned by a genetic program is divided into class boundaries using arbitrarily chosen division points; Dynamic range selection in which a subset of training examples are used to determine where, over the set of reals, class boundaries lie; Class enumeration which constructs programs similar in syntactic structure to a decision tree; and evidence accumulation which allows separate branches of the program to add to the certainty of any given class. Results showed that the dynamic range selection method was well suited to the task of multi-class classification and was capable of producing classifiers more accurate than the other methods tried when comparable training times were allowed. Accuracy of the generated classifiers was comparable to alternative approaches over several datasets.",
-
notes = "CEC-2001 - A joint meeting of the IEEE, Evolutionary
Programming Society, Galesia, and the IEE.
IEEE Catalog Number = 01TH8546C,
Library of Congress Number = .
Tested on UCI machine learning testsets. STGP. 5 approaches to multiclass classifications: binary decomposition, static range, dynamic range, class enumeration (additional data type {"}ClassType{"} (cf C4.5), evidence accumulation cf {"}AddToClass{"}, cf Teller",
