A new algorithm for non-linear mapping with applications to dimension and cluster analyses |
| |
Authors: | Osamu Kakusho Riichiro Mizoguchi |
| |
Affiliation: | The Institute of Scientific and Industrial Research, Osaka University, Suita, Osaka, 565 Japan |
| |
Abstract: | In this paper, a new non-linear mapping method suitable for dimension and cluster analysis is proposed. In order to obtain a flexible and powerful method, the non-metric multidimensional scaling of Kruskal type is extended by introducing the concept of k-nearest neighbor. Some simulation results supporting the efficiency of our new method are given along with a detailed discussion. |
| |
Keywords: | Cluster analysis Multi-dimensional scaling Minimum dimension Non-linear mapping |
本文献已被 ScienceDirect 等数据库收录! |