Issue No.03 - March (1978 vol.27)
J. Fehlauer , Drexel University
A feature extraction technique based on a new criterion for "declustering" is presented. Declustering occurs when sample vectors from one pattern class form a densely packed point constellation, or cluster, in feature space while vectors from another class do not form a cluster but instead array themselves as outliers. Features chosen to optimize the declustering criterion enhance class separation and are robust over a wide range of measurement statistics.
Rayleigh quotients, Clustering, feature extraction, linear transformation, nonparametric classification, pattern recognition, principal component analysis
J. Fehlauer, B.A. Eisenstein, "A Declustering Criterion for Feature Extraction in Pattern Recognition", IEEE Transactions on Computers, vol.27, no. 3, pp. 261-266, March 1978, doi:10.1109/TC.1978.1675083