A mixture model approach to big data clustering and classification
Hani Hamdan · 2016
In this paper, the importance and advantages of binning data, for big data clustering and classification, are shown. Then, the fundamental and basic concepts of mixture models estimation from binned data are presented. A special attention is paid to the binned-EM algorithm, and its application to data clustering and classification. A feedback on the implementation and use of this algorithm is provided. The binned-EM algorithm is summarized so that it is easy to program. In order to show the usefulness and the good performances of the presented approach to big data clustering, an example of application to image segmentation is illustrated.