A review on suppressed fuzzy c-means clustering models

Suppressed fuzzy c-means clustering was proposed as an attempt to combine the better properties of hard and fuzzy c-means clustering, namely the quicker convergence of the former and the finer partition quality of the latter. In the meantime, it became much more than that. Its competitive behavior was revealed, based on which it received two generalization schemes. It was found a close relative of the so-called fuzzy c-means algorithm with generalized improved partition, which could improve its popularity due to the existence of an objective function it optimizes. Using certain suppression rules, it was found more accurate and efficient than the conventional fuzzy c-means in several, mostly image processing applications. This paper reviews the most relevant extensions and generalizations added to the theory of fuzzy c-means clustering models with suppressed partitions, and summarizes the practical advances these algorithms can offer.