METHOD FOR COMPRESSING MULTIDIMENSIONAL GAUSSIAN DISTRIBUTION, ESPECIALLY REGARDING TO CLUSTERING PLURALITY GAUSSIAN DISTRIBUTION TO PLURALITY OF CLUSTERS TO EACH DIMENSION

摘要

PURPOSE: A method for compressing multidimensional gaussian distribution is provided to compress gaussian distribution with maintaining improved accuracy. CONSTITUTION: A first cluster is decided, selecting a centroid expressing the cluster(652). Probability distribution is designated at the cluster(606). A new centroid is calculated by considering gaussian distribution assigned to the cluster(654). The gaussian distribution is compared wit the new centroid(656). If the number of provided clusters is equal to the number of desired clusters, clustering is completed(658-660). If not, the cluster is divided into two(662).