作者: Yu-Tai Su , Chiu-Yen Tsai , Chun-Yi Sung , Cheng-Fa Tsai
DOI:
关键词:
摘要: Data clustering plays an important role in various fields. approaches have been designed recent years. This investigation aims to present data algorithm identify potential musical instruments teachers. With a total of 5125 candidates registered respectively 9 grades Taiwan United Music Grade Test during 2000-2008. Moreover, this study proposes new called MIDBSCAN and existing well-known neural network self-organizing map (SOM) perform applications for discovering The processing procedure searching neighbors (neighborhood points) is very time consuming the DBSCAN IDSCAN algorithms. Therefore, shorten consumed, proposed focuses lowering number expansion seeds added into neighborhood procedure, thus reducing cost neighbors. According our simulation results, approach has low execution cost, maximum deviation correctness rate noise filtering rate. outperforms SOM cost. It feasible analysis mining using algorithm.