Prof. Emo Welzl and Prof. Bernd Gärtner
|Mittagsseminar Talk Information|
Date and Time: Thursday, April 22, 2004, 12:15 pm
Duration: This information is not available in the database
Location: This information is not available in the database
Speaker: Simon Spalinger
An approach for clustering that uses support vector machines is presented. Data points are mapped by means of a Gaussian Kernel to a high dimensional feature space, where the minimum enclosing sphere is found. This sphere, when mapped back to input space, can seperate into several componenents. The components contours are interpreted as cluster boundaries.
In the second part of the talk we will explore the idea of using the support vector approach for surface reconstruction: can the contours of the feature space sphere be used to implicitly define surfaces in three dimensional input space?
(A. Ben-Hur, D. Horn, H. Siegelmann, V. Vapnik, Jorunal of Machine Learning Research 2 (2001), pp. 125-137)
Automatic MiSe System Software Version 1.4803M | admin login