Title :
An inheritable clustering algorithm suited for parameter changing
Author :
Li, Fei ; Liu, Shang ; Dou, Zhi-Tong ; Huang, Ya-lou
Author_Institution :
Lab. of Intelligent Inf. Process., NanKai Univ., Tianjin, China
Abstract :
DBSCAN is a classic density based algorithm and it clusters the data set according to the user input parameters. This work investigates how to inherit the mining results of last time when parameters change. A new incremental clustering algorithm IPC-DBSCAN is proposed, which gets the same result as that of rerunning DBSCAN yet high efficiency is achieved. Theoretical analysis and experimental results show that the proposed method reduces search space greatly and has novel efficiency. By interaction, IPC-DBSCAN gets the most satisfying result quickly and especially suits large volume data set.
Keywords :
data mining; pattern clustering; statistical analysis; IPC-DBSCAN; data mining; incremental clustering algorithm; inheritable clustering algorithm; Association rules; Clustering algorithms; Data mining; Educational institutions; Filtering algorithms; Information processing; Information science; Laboratories; Machine learning algorithms; Partitioning algorithms;
Conference_Titel :
Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference on
Print_ISBN :
0-7803-8403-2
DOI :
10.1109/ICMLC.2004.1382373
