Learning metrics for information visualization

Reference:

Jaakko Peltonen, Arto Klami, and Samuel Kaski. Learning metrics for information visualization. In Proceedings of WSOM'03, Workshop on Self-Organizing Maps, pages 213–218, Kitakyushu, Japan, 2003. Kyushu Institute of Technology. (Proceedings on CD-ROM).

Abstract:

The learning metrics principle shows how (nonlinear) projection and clustering methods can be made to focus on discriminative properties of data. In this paper we review and extend our earlier work on learning metrics for self-organizing maps (SOMs), compare algorithms, and introduce a new accurate distance computation algorithm. It can be used with methods that work on pairwise distances between the data samples. Its usefulness is demonstrated for Sammon's mapping, a form of multidimensional scaling.

Suggested BibTeX entry:

@inproceedings{Peltonen03wsom,
    address = {Kitakyushu, Japan},
    author = {Jaakko Peltonen and Arto Klami and Samuel Kaski},
    booktitle = {Proceedings of WSOM'03, Workshop on Self-Organizing Maps},
    note = {(Proceedings on CD-ROM)},
    pages = {213-218},
    publisher = {Kyushu Institute of Technology},
    title = {Learning metrics for information visualization},
    year = {2003},
}