How many neurons?: a genetic programming answer

Published in Conferences Papers
  1. Leonardo Trujillo, Yuliana Mart\'ınez and Patricia Melin. Estimating Classifier Performance with Genetic Programming. In Proceedings of the 14th European Conference on Genetic Programming. 2011, 274–285. URL BibTeX

    @inproceedings{Trujillo:2011:ECP:2008307.2008333,
    	author = "Trujillo, Leonardo and Mart\'{\i}nez, Yuliana and Melin, Patricia",
    	title = "Estimating Classifier Performance with Genetic Programming",
    	booktitle = "Proceedings of the 14th European Conference on Genetic Programming",
    	series = "EuroGP'11",
    	year = 2011,
    	isbn = "978-3-642-20406-7",
    	location = "Torino, Italy",
    	pages = "274--285",
    	numpages = 12,
    	url = "http://dl.acm.org/citation.cfm?id=2008307.2008333",
    	acmid = 2008333,
    	publisher = "Springer-Verlag",
    	address = "Berlin, Heidelberg"
    }
    
Abstract

The goal of this paper is to derive predictive models that take as input a description of a problem and produce as output an estimate of the optimal number of hidden nodes in an Artificial Neural Network (ANN). We call such computational tools Direct Estimators of Neural Network Topology (DENNT), an use Genetic Programming (GP) to evolve them. The evolved DENNTs take as input statistical and complexity descriptors of the problem data, and output an estimate of the optimal number of hidden neurons.

Published in
GECCO '11 Proceedings of the 13th annual conference companion on Genetic and evolutionary computation
Pages 175-176
http://dl.acm.org/citation.cfm?id=2001956&dl=ACM&coll=DL&CFID=249301493&CFTOKEN=11015299
Date of conference
12 - 16 July 2011
ISBN
978-1-4503-0690-4
Feedback