Title :
Improving the running time of the nearest neighbor algorithm
Author :
Chompupatipong, Nattakon ; Jearanaitanakij, Kietikul
Author_Institution :
Dept. of Comput. Eng., King Mongkut´s Inst. of Technol. Ladkrabang, Bangkok, Thailand
Abstract :
Nearest neighbor algorithm is a well-known method, in pattern recognition, for classifying objects based on the nearest examples in the feature space. However, it´s major drawback is the sequential search operation which calculates the distance between the probing object and the entire set of the training instances. In this paper, we propose a novel method to accelerate the searching operation in the nearest neighbor algorithm. Our method consists of two main steps; creating the reference table and searching the nearest neighbor. Reference table of the training instances is created once in the initial phase and referred periodically by the searching step. Surprisingly, this reference table can drastically reduce the searching time of the nearest neighbor algorithm on any feature space. The experimental results on five real-world datasets from the VCI repository show a remarkable improvement on the searching time while the accuracy is still preserved.
Keywords :
learning (artificial intelligence); pattern classification; VCI repository; feature space; nearest neighbor algorithm; nearest neighbor search; object classification; pattern recognition; reference table creation; training instances; Accuracy; Classification algorithms; Computer science; Euclidean distance; Sonar; Time complexity; Training; Euclidean Distance; Geometric Measurement; Nearest Neighbor Algorithm; Reference Table; Running Time;
Conference_Titel :
Computer Science and Engineering Conference (ICSEC), 2013 International
Conference_Location :
Nakorn Pathom
Print_ISBN :
978-1-4673-5322-9
DOI :
10.1109/ICSEC.2013.6694795