Incremental algorithms for fuzzy co-clustering of very large cooccurrence matrix
Katsuhiro Honda, Daiji Tanaka, Akira Notsu · 2014
Handling very large data is an important issue in FCM-type clustering and several incremental algorithms have been proved to be useful in FCM clustering. In this paper, the incremental algorithms are extended to fuzzy co-clustering of cooccurrence matrices, whose goal is to simultaneously partition objects and items considering their cooccurrence information. Single pass and online approaches are applied to fuzzy clustering for categorical multivariate data (FCCM) and fuzzy CoDoK, which try to maximize the aggregation degrees of co-clusters adopting entropy-based and quadratic-based membership fuzziflcations. Several experimental results demonstrate the applicability of the incremental approaches to fuzzy co-clustering algorithms.