著者

古川 徹生

出版者

日本知能情報ファジィ学会

雑誌

知能と情報 : 日本知能情報ファジィ学会誌 : journal of Japan Society for Fuzzy Theory and Intelligent Informatics (ISSN:13477986)

巻号頁・発行日

vol.19, no.6, pp.618-626, 2007-12-15

参考文献数

7

被引用文献数

1

---

This paper introduces an extension of an SOM called the "SOM of SOMs," or SOM^2, in which objects to be mapped are self-organizing maps. In SOM^2, each nodal unit of a conventional SOM is replaced by a function module of SOM. Since each child SOM module in SOM^2 is trained to represent an individual map, the parent map in SOM^2 generates a self-organizing map representing the continuous change of the child maps. Thus SOM^2 is an extension from "self-organizing map" to "self-organizing homotopy". From another viewpoint, SOM^2 is a learning machine which represents a fiber bundle, whereas the conventional SOM represents a manifold. This paper presents the architecture and the algorithm of SOM^2 as well as some application results.