The Construction of Cascaded Features and Its Application in Fuzzy Clustering

Abstract

The success of fuzzy clustering heavily relies on the proper feature space constructed by the input data. For nonspherical and overlapped clusters, kernel fuzzy clustering is more effective owing to it finds more proper feature space compared to conventional fuzzy clustering. Unfortunately, poor scalability of kernel fuzzy clustering is induced by the requirement of large memory and running time. To solve the problem, random feature based method was presented to approximate the kernel function. Features in this approximate feature space are very useful information. Inspired by the architecture of functional-link neural network, to represent the diversity of features, cascaded features are constructed by a new feature mapping technique called cascaded feature mapping in this paper. By performing classical fuzzy c-means (FCM) with the cascaded features, a new fuzzy clustering algorithm called FCM-CF is developed. The experimental results of our proposed methods verify the superiority in comparison of other classical fuzzy clustering methods.