Collision Description of Clustering
2009
Clustering is the hot spot in the study of Data Mining. In this paper, refer to the distillation of "collision", clustering changes can be described by energy and collision. The process optimized by energy and collision has the energy of user’s theme demands drive on the data space of initial clustering form. The energy makes data-objects continuously hit with each other. In the process of collision, the form of collision is changing with the user’s theme demands and to describe the incidence relation between clusters. Meanwhile, in the process, the transaction of data-objects in the outlier is also considered during the process of collision. According to the energy driven by user’s theme demands, the data-objects in the cluster can transform to the data-object in the outlier. Or the data-objects in the outlier can also join some cluster. Through the experiment, a strategy of clustering optimized by energy and collision is better than those which are just clustered by using the CRCMD algorithm based on density.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
14
References
0
Citations
NaN
KQI