Visualization of high-dimensional data is an important issue in data mining as it enhances the chance to selectively choose appropriate techniques for analyzing data. In this paper, two extensions to recent angle based multi-dimensional scaling techniques are presented. The first approach concerns the preprocessing of the data with the objective to lower the error of the subsequent mapping. The second aims at improving differentiability of angle based mappings by augmenting the target space by one additional dimension. Experimental results demonstrate the gain of efficiency in terms of layout quality and computational complexity. © 2008 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Rehm, F., & Klawonn, F. (2008). Improving angle based mappings. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 5139 LNAI, pp. 3–14). Springer Verlag. https://doi.org/10.1007/978-3-540-88192-6_3
Mendeley helps you to discover research relevant for your work.