Loading...
Please wait, while we are loading the content...
Similar Documents
Efficient density clustering using basin spanning trees
| Content Provider | Semantic Scholar |
|---|---|
| Author | Hader, Sören Hamprecht, Fred A. |
| Copyright Year | 2003 |
| Abstract | We present a method to cluster multivariate data according to their density (or any other target function): all observations lying within one “basin” or sitting on the slopes of one “mountain” are assigned to one cluster. The method exploits a neighborhood structure given by the Delaunay triangulation or a k-nearest neighbor graph, and each cluster is given in terms of a basin spanning tree. The root of each basin spanning tree corresponds to a local density maximum, and the trees can be used for simplified representation and visualization of the observations. We compare the accuracy and speed of different approximations, apply the method to real-world data sets and compare its computational complexity to published algorithms. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://klimt.iwr.uni-heidelberg.de/mip/fhamprecht/efficient-density-clustering.pdf |
| Language | English |
| Access Restriction | Open |
| Subject Keyword | Ability to Sit question Approximation Computational complexity theory Delaunay triangulation File spanning Graph - visual representation Imagery K-nearest neighbors algorithm PersonNameUse - assigned Scientific Publication Single Linkage Cluster Analysis Spanning tree Telling untruths Trees (plant) slope statistical cluster |
| Content Type | Text |
| Resource Type | Article |
