An Expanding Clustering Algorithm Based on Density Searching
Most clustering algorithms need to preset the initial parameters which affect the performance of clustering very much. To solve this problem, a new method is proposed, which determine the center points of clustering by density-searching according to the universality of the Gaussian distribution. After the center was obtained, the cluster expands based on the correlation coefficient between clusters and the membership of the samples until the terminating condition is met. The experimental results show that this method could classify the samples of Gaussian distribution with different degree of overlap accurately. Compared with the fuzzy c-means algorithm, the proposed method is more accurate and timesaving when applied to the Iris data and Fossil data.
Keywordsclustering density searching clustering center Algorithm
Unable to display preview. Download preview PDF.
- 1.Hu, B.-p., He, X.-s.: Novel BSS algorithm for estimating PDF based on SVM. Computer Engineering and Applications 45(17), 142–144 (2009)Google Scholar
- 3.Yu, F.-x., Su, J.-y., Lu, Z.-m., et al.: Multi-feature based fire detection in video. International Journal of Innovative Computing, Information and Control 4(8), 1987–1993 (2008)Google Scholar
- 6.Zhou, X.-Y., Zhang, J., Sun, Z.-H.: An Efficient Clustering Algorithm for High Dimensional Turnstile Data Streams. Computer Science 33(11), 14–17 (2006)Google Scholar