Summary
International Technical Conference on Circuits/Systems, Computers and Communications
2008
Session Number:P1
Session:
Number:P1-31
Optimization of Distances for a Stochastic Embedding and Clustering of High-Dimensional Data
Naoto Nishikawa, Shinji Doi,
pp.-
Publication Date:2008/7/7
Online ISSN:2188-5079
DOI:10.34385/proc.39.P1-31
PDF download (315.9KB)
Summary:
The stochastic proximity embedding (SPE) is a method of data visualization in research area of data clustering and mining. The SPE can visualize high-dimensional data by embedding them in a low-dimensional space according to a given similarity among input data. This paper extends the SPE by applying a simple iterative learning process. Without any knowledge on data, the extended SPE can automatically optimize the similarity of data and can produce low-dimensional embeddings more accurately than the original SPE.