Issue No. 12 - December (2003 vol. 25)
Sameer Singh , IEEE
<p><b>Abstract</b>—In this paper, we study two measures of classification complexity based on feature space partitioning: "purity" and "neighborhood separability." The new measures of complexity are compared with probabilistic distance measures and a number of other nonparametric estimates of classification complexity on a total of 10 databases from the University of Calfornia, Irvine, (UCI) repository.</p>
Pattern recognition, classification complexity, feature space partitioning.
S. Singh, "Multiresolution Estimates of Classification Complexity," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 25, no. , pp. 1534-1539, 2003.