A Study Of The Pattern-Based Clustering Theories

For DNA microarry analysis, traditional method that worked out the similarity by distance is not so proper because sometimes there doesn't exist approximate distance but a coherent pattern. Therefore, a brand-new clustering model : Pattern-Based Clustering, was proposed to slove this problem. Two objects in the same cluster is decided whether the subsets they belong to share a coherent pattern. Recently the study of Pattern-based Clustering is merely have ecumenical descriptions but didn't have system info and lacks proper definitions. Furthermore, current clustering algorithms mostly based on specific shifting or scaling patterns, but can not capture other patterns properly.This paper makes deep study and analysis of different pattern-based clustering models, summarizes their commonness and differences. For the study of theory, first of all, makes multiformity analysis of different patterns and their expression feathers, then proposed the definitions of single-pattern models and multi-pattern models. From the commonness and differences of patterns, the common definition and formula description of pattern are summarized. Then a general symmetrical arithmetic operator is proposed. Based on properties of this arithmetic operator, a study of clustering operate principle is processed. At last, For the study of algorithm, we purposed a new pattern-based algorithm called 0-SM, which is more efficient and precise than prior approaches.