The Shrinking-clustering Method and Simulation to High Dimensional Data
A shrinking clustering method is brought out to solve the clustering problem of high dimensional data in data mining. The concept of the variable grid and the technique of multi-scale grid recognition are presented to clarify the quantization value of data structure. And the clustering rule of different scale had also been clarified. To upgrade the conventional method of artificial cluster-detection, the automatically shrinking model is proposed as well as the clustering with noise. Furthermore, simulation result based on integrated- movement of data and grid compression algorithm shows that the method can detect clusters effectively and efficiently in both low and high dimensional data.
Jian-ye Zhang Quan Pan Jian-hai Liang
School of Automation, Northwestern Polytechnical University Xian, ST 710072 China Institute of Engineering, Air Force Engineering University Xian, ST 710038 China
国际会议
北京
英文
2007-08-05（万方平台首次上网日期，不代表论文的发表时间）