Title :
Multiresolution estimates of classification complexity
Author_Institution :
Dept. of Comput. Sci., Exeter Univ., UK
Abstract :
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 California, Irvine, (UCI) repository.
Keywords :
computational complexity; decision trees; image resolution; nonparametric statistics; pattern classification; probability; Irvine; University of California; classification complexity; decision trees; feature space partitioning; multiresolution estimates; neighborhood separability; nonparametric estimates; probabilistic distance measures; Decision trees; Entropy; Error analysis; Extraterrestrial measurements; Impurities; Partitioning algorithms; Pattern analysis; Pattern recognition; Spatial databases; Testing;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on
DOI :
10.1109/TPAMI.2003.1251146
