Fuzzy Clustering and Superclustering Scheme for Extracting Structure from Data

摘要

A fuzzy clustering algorithm based on a least-square cost function is examined.Given inorganic sensor data, the algorithm can without operator intervention decompose the data into clusters associated with particular emitters. The algorithm determines a fuzzy cluster center that represents a reduced noise estimate of measured quantities. Also, the algorithm finds the grade of membership of each data point in each cluster. The grade of membership tells the degree to which each data point belongs to a cluster, and it can be used as a measure of confidence in the cluster assignment. A second component of the algorithm superclustering allows the number of targets present in the data to be determined without a priori information, also allowing a refinement of the fuzzy cluster centers and grades of membership. The combined fuzzy clustering and superclustering algorithm is applied to both simulated and electronic support measures (ESM) data.