On the selection of base prototypes for LAESA and TLAESA classifiers

—The k nearest neighbor (k-NN) classifier has been extensively used as a nonparametric technique in Pattern Recognition. However, in some applications where the training set is large, the exhaustive k-NN classifier becomes impractical. To avoid this problem, many fast k-NN classifiers have been developed. AESA is one of the most popular fast k-NN classifiers, due to its good behavior. This classifier requires storing all distances between every pair of prototypes in the training set, which is a drawback, particularly when the training set is large. For this reason, in order to reduce the space requirements, some improvements such as LAESA and TLAESA have been proposed. One important step for the performance of LAESA and TLAESA is the base prototypes (BP) selection algorithm. In this work, two BP selection algorithms are proposed (BPClass and BPRepProt) and evaluated over LAESA and TLAESA classifiers. Some experiments with public standard numerical datasets, are presented. Keywords-fa...

• Type: Conference
• Year: 2008
• Where: ICPR
• Authors: Selene Hernández-Rodríguez, José Francisco Martínez Trinidad, Jesús Ariel Carrasco-Ochoa
• Keywords: Computer Vision, Fast K-NN Classifiers, ICPR 2008, K-nn Classifiers, Training Set