A simple "possibilistic" clustering neural network
Orly Yadid-Pecht, Moshe Gur · 2002
A simple "possibilistic" clustering method i.e. clustering where each datum has a degree of possibility of belonging to the cluster, using a neural net, is suggested. The implementation consists of simple "neurons", requiring only a small number of local connections, collectively performing a diffusion-like process. In spite of its simplicity, this implementation has several advantages over commonly used fuzzy clustering methods. Specifically, it provides the "typicality" notion that is lacking in the well known Fuzzy C Means (FCM) and its derivatives, and is less sensitive to noise.