- author = "Martijn C. J. Bot",
- title = "Feature Extraction for the k-Nearest Neighbour Classifier with Genetic Programming",
- booktitle = "Genetic Programming, Proceedings of EuroGP'2001",
- year = "2001",
- editor = "Julian F. Miller and Marco Tomassini and Pier Luca Lanzi and Conor Ryan and Andrea G. B. Tettamanzi and William B. Langdon",
- volume = "2038",
- series = "LNCS",
- pages = "256--267",
- address = "Lake Como, Italy",
- publisher_address = "Berlin",
- month = "18-20 " # apr,
- organisation = "EvoNET",
- publisher = "Springer-Verlag",
- keywords = "genetic algorithms, genetic programming, Feature Extraction, Machine Learning: Poster",
- ISBN = "3-540-41899-7",
-
DOI = "
doi:10.1007/3-540-45355-5_20",
- size = "12 pages",
-
abstract = "In pattern recognition the curse of dimensionality can
be handled either by reducing the number of features,
e.g. with decision trees or by extraction of new
features.
We propose a genetic programming (GP) framework for automatic extraction of features with the express aim of dimension reduction and the additional aim of improving accuracy of the k-nearest neighbour (k-NN) classifier. We will show that our system is capable of reducing most datasets to one or two features while k-NN accuracy improves or stays the same. Such a small number of features has the great advantage of allowing visual inspection of the dataset in a two-dimensional plot.
Since k-NN is a non-linear classification algorithm, we compare several linear fitness measures. We will show the a very simple one, the accuracy of the minimal distance to means (mdm) classifier outperforms all other fitness measures.
We introduce a stopping criterion gleaned from numeric mathematics. New features are only added if the relative increase in training accuracy is more than a constant d, for the mdm classifier estimated to be 3.3%.",
- notes = "EuroGP'2001, part of \cite{miller:2001:gp}",
