Majority-voting FCM algorithm in the vague fuzzy classification

The FCM algorithm is an essential tool for the fuzzy classification. But it suffers vague decision diversity due to the heavy dependence upon the initial randomization. To provide less vague decision strategy we suggest a modified FCM algorithm that uses multiple randomizations of the membership functions. The algorithm inherently provides the parallelism and implied stochastic annealing of the convergence trajectory. The strategy to switch among multiple prototypes during each simulation step is the "majority voting" in nature. The algorithm naturally provides the "averaged rate of convergence" and a convincing decision measure to accept a "more suitable" class among huge set of all legitimate classifications. The contributions are 1) the MV FCM algorithm, 2) the sketch of the convergence analysis.