Parallel attribute-weighted fuzzy c-means algorithm for clustering

Due to energy and bandwidth constraints, traditional data clustering approaches are challenging when the communication to a central processing unit is discouraged in Wireless Sensor Networks. In this paper, a new parallel attribute-weighted fuzzy c-means (PWFCM) algorithm is proposed, in which parallel clustering solution can be achieved by exchanging the centroid messages among single-hop neighbours only. At the same time, the important features can be extracted based on the attribute weight entropy regularization. Experiments on real and synthetic datasets have demonstrated the suitability and efficiency of the presented algorithm.