The Nuclear Clustering Algorithm of a Class of Metric Space
Zhu Jian-Guo, Yu Kai-Rong, Zhu Xiang-wei · 2012
Through a number of lemma and theorems, nuclear clustering method proposed in this paper solves a class of space-metric nuclear clustering, which uses matrix elementary transformations and adopts a nonlinear mapping to extract differentiate and amplify characteristics of useful information. This method strives to avoid clustering that uses the characteristics of the sample directly. Even though the spread of the sample is uneven or the sample distribution is chaos, nuclear clustering method can achieve accurate clustering. This paper also gives the algorithm of nuclear clustering.