Fast k most similar neighbor classifier for mixed data based on approximating and eliminating

摘要: The k nearest neighbor (k-NN) classifier has been a widely used nonparametric technique in Pattern Recognition. In order to decide the class of new prototype, k-NN performs an exhaustive comparison between prototype classify (query) and prototypes training set T. However, when T is large, expensive. To avoid this problem, many fast algorithms have developed. Some these are based on Approximating-Eliminating search. case, Approximating Eliminating steps rely triangle inequality. soft sciences, usually described by qualitative quantitative features (mixed data), sometimes function does not satisfy Therefore, work, most similar neighbour for mixed data (AEMD) presented. This consists two phases. first phase, binary similarity matrix among stored. second steps, which inequality, proposed compared against other algorithms, adapted work with data. experiments real datasets presented