Basic Consideration of Online and Mini-Batch Algorithms for MMMs-induced Fuzzy Co-clustering
Seiki Ubukata, Akira Notsu, Keiko Kida, Katsuhiro Honda · 2018
Fuzzy co-clustering schemes including Fuzzy Co-Clustering induced by Multinomial Mixture models (FCCMM) are promising approaches for analyzing object-item cooccurrence information such as document-keyword frequencies and customer-product purchase history transactions. However, such cooccurrence datasets are generally maintained as very large matrices and cannot be dealt with conventional batch algorithms. Online algorithms that load sequentially a single object for adjusting parameters are effective approaches for big data analysis. Mini-batch algorithms that load sequentially a small chunk (mini-batch) of objects for adjusting parameters are also effective. In this paper, we propose an online algorithm for FCCMM clustering and a mini-batch algorithm for FCCMM clustering and observe their characteristics and performance through numerical experiments.