Abstract | ||
---|---|---|
Grouping data into meaningful clusters is an important task of both artificial intelligence and data mining. An important group of clustering algorithms are density based ones that require calculation of a neighborhood of a given data point. The bottleneck for such algorithms are high dimensional data. In this paper, we propose a new TI-k-Neighborhood-Index algorithm that calculates k-neighborhoods for all points in a given data set by means the triangle inequality. We prove experimentally that the NBC (Neighborhood Based Clustering) clustering algorithm supported by our index outperforms NBC supported by such known spatial indices as VA-file and R-tree both in the case of low and high dimensional data. |
Year | DOI | Venue |
---|---|---|
2010 | 10.1007/978-3-642-15381-5_35 | IDEAL |
Keywords | Field | DocType |
triangle inequality,neighborhood-based clustering,artificial intelligence,important task,important group,data mining,data point,known spatial index,grouping data,new ti-k-neighborhood-index algorithm,clustering algorithm,high dimensional data,grouped data,artificial intelligent,indexation | Data mining,Canopy clustering algorithm,CURE data clustering algorithm,Data stream clustering,Correlation clustering,Computer science,Constrained clustering,FLAME clustering,Artificial intelligence,Cluster analysis,Machine learning,Single-linkage clustering | Conference |
Volume | ISSN | ISBN |
6283 | 0302-9743 | 3-642-15380-1 |
Citations | PageRank | References |
7 | 0.60 | 5 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Marzena Kryszkiewicz | 1 | 1662 | 118.72 |
Piotr Lasek | 2 | 64 | 4.15 |