Wrappers for Feature Subset Selection
		    Ron Kohavi and George H. John
		     Computer Science Department
			 Stanford University
			  Stanford, CA 94305
		    {ronnyk,gjohn}@CS.Stanford.EDU
	     http://robotics.stanford.edu/~{ronnyk,gjohn}

  In the feature subset selection problem, a learning algorithm is
  faced with the problem of selecting a relevant subset of features
  upon which to focus its attention, while ignoring the rest.  To
  achieve the best possible performance with a particular learning
  algorithm on a particular domain, a feature subset selection method
  should consider how the algorithm and the training data interact.
  We explore the relation between optimal feature subset selection and
  relevance.  Our wrapper method searches for an optimal feature
  subset tailored to a particular algorithm and a domain.  We study
  the strengths and weaknesses of the wrapper approach and show
  improvements over the original design.  We compare the wrapper
  approach to induction without feature subset selection and to
  Relief, a filter-based approach to feature subset selection.
  Significant improvement in accuracy on real problems is achieved for
  the two families of induction algorithms used: decision trees and
  Naive-Bayes.

Citation:
Kohavi, R. and John, G. H. (1997), Wrappers for Feature Subset
Selection.  _Artificial Intelligence Journal_.  Forthcoming.