Improved FCM Algorithm Based on the Density of Samples
Feng-feng Zhu · Science Technology and Engineering · 2007
By the intuitionist property of cluster centroid,the data points of greater density are selected as the initial centroid of clusters.With this improvement,the algorithm has overcome the typical FCM drawbacks such as the sensitivity to initialization and the tendency to get trapped in local minima.Experimental results have testified the feasibility and validity of the algorithm.