Fuzzy clustering with hedge algebra
Đinh Khắc Đông, Tran Dinh Khang, Phan Anh Phong · 2010
In this paper, we propose a new approach to fuzzy clustering in order to handle the uncertainties in pattern recognition problems on the basis of conventional fuzzy C-means algorithm (FCM). In our approach, we define the concept of linguistic cluster center by employing the semantic structure of hedge algebra. This kind of cluster center is constructed to give the appropriate weights for each pattern of the dataset in our clustering algorithm. The parameters of hedge algbra are then optimized in the training process to obtain the suitable parameters for the dataset. We also incorporate the k-means algorithm to get better results in comparing to conventional FCM.