Adaptive Algorithms for Weight of the Feature Weighted of FCM

On the base of FCM arithmetic which is in existence, in order to get the preferable separating effect, we bring forward an adaptive algorithm for the weight of the feature weighted of FCM which founds on the feature contribution balance principle and the most separate degree principle of intra-cluster. The arithmetic avoids each feature of the feature data originated can´t be compared non-comparatives of each feature of the feature vector originated by adopting different units, and ignores the features whose contribution is small for clustering, and reduces the complicacy of the calculation. And by the simulation of the IRIS, we find the calculation method for weight is efficiency.