Fuzzy clustering with the generalized entropy of feature weights
Kai Li, Yan Gao · International Journal of Advanced Computer Research · 2016
Clustering based on objective function is a commonly used method, which is attributed as an optimization problem with the constrained conditions, such as kmeans clustering.As this method has some flaws, the researchers conduct some improvements on k-means clustering algorithm.It is worth mentioning that the researchers introduced the feature weights into objective function of optimization problems involved and presented a lot of different clustering algorithms.For example, Huang et al. [1] introduced feature weights into the objective function of k-means clustering and proposed WK-Means clustering algorithm.Renato et al. [2] further extended the Euclidean distance to Minkowski distance and studied the relationship between feature weights and measure of distance.In addition, an improved method for k-means clustering algorithm or its variants is proposed to the objective function of optimization problem.Since then, researchers presented fuzzy cmeans (FCM) clustering algorithms and its variants [3-5].Some researchers proposed the maximum entropy clustering algorithm in order to overcome the deficiencies of FCM [6, 7].