An improved robust fuzzy clustering algorithm

Fuzzy-c-means (FCM) and hard clustering algorithms are the most common tools for data partitioning. However, these clustering algorithms may fail completely in the presence of noise. We introduce a robust noise rejection clustering algorithm based on a combination of techniques that treat the FCM weak points with a traditional noise rejection algorithm. Unlike the traditional FCM, the proposed algorithm is a powerful tool for partitioning the data in the presence of noise (outliers).