Browse by topic
Type of publication
Improving classification using median string and NN rules. Proc. of the IX Spanish Symposium on Pattern Recognition and Image Analysis, 2001. pp. 391-395.Improving classification using median string and NN rules Carlos-D. Martínez-Hinarejos, Alfons Juan, Francisco Casacuberta In Pattern Recognition, the concept of (generalized) median string is important for modelling a (large) set of garbled strings or patterns. The search for such a string is a difficult computational problem; thus, only suboptimal approaches of median string can be computed with a reasonable effort. Recently, a greedy algorithm was proposed to compute an approximate median string of a set of strings using an iterative improvement. In this work, we propose an alternative definition of median string which is introduced into this algorithm. Experiments have been carried out with real data to compare the performances of the Nearest-Neighbours classifiers based on set median and median strings.These experiments showed that the new definition of median string gives better results than obtained with set median and classical median strings.