Title :
Feature selection for multiclass discrimination via mixed-integer linear programming
Author :
Iannarilli, Frank J., Jr. ; Rubin, Paul A.
Author_Institution :
Aerodyne Res. Inc., Billerica, MA, USA
fDate :
6/1/2003 12:00:00 AM
Abstract :
We reformulate branch-and-bound feature selection employing L∞ or particular Lp metrics, as mixed-integer linear programming (MILP) problems, affording convenience of widely available MILP solvers. These formulations offer direct influence over individual pairwise interclass margins, which is useful for feature selection in multiclass settings.
Keywords :
feature extraction; integer programming; linear programming; pattern classification; branch-and-bound; feature selection; mixed-integer linear programming; pairwise interclass margins; pattern classification; pattern recognition; Covariance matrix; Employment; Gaussian distribution; Gaussian processes; Linear programming; Pattern classification; Pattern recognition; Q measurement; Statistical distributions;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on
DOI :
10.1109/TPAMI.2003.1201827
