Nate Kohl's Publications

Sorted by DateClassified by Publication TypeClassified by Research Category


Automatic Feature Selection in Neuroevolution

Automatic Feature Selection in Neuroevolution. Shimon Whiteson, Peter Stone, Kenneth O. Stanley, Risto Miikkulainen, and Nate Kohl. In GECCO 2005: Proceedings of the Genetic and Evolutionary Computation Conference, pp. 1225–1232, June 2005.
http://www.isgec.org/gecco-2005/

Download

[postscript]1.6MB  

Abstract

Feature selection is the process of finding the set of inputs to a machine learning algorithm that will yield the best performance. Developing a way to solve this problem automatically would make current machine learning methods much more useful. Previous efforts to automate feature selection rely on expensive meta-learning or are applicable only when labeled training data is available. This paper presents a novel method called FS-NEAT which extends the NEAT neuroevolution method to automatically determine an appropriate set of inputs for the networks it evolves. By learning the network's inputs, topology, and weights simultaneously, FS-NEAT addresses the feature selection problem without relying on meta-learning or labeled data. Initial experiments in an autonomous car racing simulation demonstrate that FS-NEAT can learn better and faster than regular NEAT. In addition, the networks it evolves are smaller and require fewer inputs. Furthermore, FS-NEAT's performance remains robust even as the feature selection task it faces is made increasingly difficult.

BibTeX Entry

@InProceedings{whiteson:gecco05,
  author = "Shimon Whiteson and Peter Stone and Kenneth O. Stanley and Risto Miikkulainen and Nate Kohl",
  title = "Automatic Feature Selection in Neuroevolution",
  booktitle = "GECCO 2005: Proceedings of the Genetic and Evolutionary Computation Conference",
  month = "June",
  year = "2005",
  pages = "1225--1232",
  abstract = {
  Feature selection is the process of finding the set of inputs to a
  machine learning algorithm that will yield the best performance.
  Developing a way to solve this problem automatically would make
  current machine learning methods much more useful.  Previous efforts
  to automate feature selection rely on expensive meta-learning or are
  applicable only when labeled training data is available.  This paper
  presents a novel method called FS-NEAT which extends the NEAT
  neuroevolution method to automatically determine an appropriate set
  of inputs for the networks it evolves.  By learning the network's
  inputs, topology, and weights simultaneously, FS-NEAT addresses the
  feature selection problem without relying on meta-learning or
  labeled data.  Initial experiments in an autonomous car racing
  simulation demonstrate that FS-NEAT can learn better and faster than
  regular NEAT.  In addition, the networks it evolves are smaller and
  require fewer inputs.  Furthermore, FS-NEAT's performance remains
  robust even as the feature selection task it faces is made
  increasingly difficult.
  }
  wwwnote = {<a href="http://www.isgec.org/gecco-2005/">http://www.isgec.org/gecco-2005/</a>},
  bib2html_pubtype = {Refereed Conference},
  bib2html_rescat = {Machine Learning}
}

Generated by bib2html.pl (written by Patrick Riley ) on Sun May 15, 2011 08:27:15