Regularized linear fuzzy clustering and probabilistic PCA mixture models

摘要: Fuzzy c-means (FCM)-type fuzzy clustering approaches are closely related to Gaussian mixture models (GMMs) and EM-like algorithms have been used in FCM with regularized objective functions. Especially, regularization by Kullback-Leibler information (KLFCM) is a counterpart of GMMs. In this paper, we propose apply probabilistic principal component analysis (PCA) linear following discussion on the relationship between local PCA clustering. Although proposed method kind constrained model KLFCM, algorithm includes c-varieties (FCV) as special case, can be regarded modified FCV K-L information. Numerical experiments demonstrate that more flexible than maximum likelihood useful for capturing substructures properly.