The Fuzzy Clustering Problem (FCP) is a mathematical program which is
difficult to solve since it is nonconvex, which implies possession of
many local minima. The fuzzy C-means heuristic is the widely known app
roach to this problem, but it is guaranteed only to yield local minima
. In this paper, we propose a new approach to this problem which is ba
sed on tabu search technique, and aims at finding a global solution of
FCP. We compare the performance of the algorithm with the fuzzy C-mea
ns algorithm. (C) 1997 Pattern Recognition Society. Published by Elsev
ier Science Ltd.