- title = "Automatically Defined Features: The Simultaneous Evolution of 2-Dimensional Feature Detectors and an Algorithm for Using Them",
- author = "David Andre",
- booktitle = "Advances in Genetic Programming",
- publisher = "MIT Press",
- editor = "Kenneth E. {Kinnear, Jr.}",
- year = "1994",
- pages = "477--494",
- chapter = "23",
- keywords = "genetic algorithms, genetic programming",
-
URL = "
http://www.amazon.co.uk/Advances-Genetic-Programming-Complex-Adaptive/dp/0262111888",
-
URL = "
http://cognet.mit.edu/sites/default/files/books/9780262277181/pdfs/9780262277181_chap23.pdf",
-
DOI = "
doi:10.7551/mitpress/1108.003.0029",
- size = "18 pages",
- abstract = "Although automatically defined functions (ADFcts) with genetic programming (GP) appear to have great utility in a wide variety of domains, their application to the automatic discovery of 2-dimensional features has been only moderately successful [Koza 1993]. Boolean functions of pixel inputs, although very general, may not be the best representation for 2-dimensional features. This chapter describes a method for the simultaneous evolution of 2-dimensional hit-miss matrices and an algorithm to use these matrices in pattern recognition. Hit-miss matrices are templates that can be moved over part of an input pattern to check for a match. These matrices are evolved using a 2-dimensional genetic algorithm, while the algorithms controlling the templates are evolved using GP. The approach is applied to the problem of digit recognition, and is found to be successful at discovering individuals which can recognize very low resolution digits. Possibilities for expansion into a full-size character recognition system are discussed.",
-
notes = "
Mixture of GP and two dee GA
Part of \cite{kinnear:book}",
