INB Lunch-Seminar

Extensions of self-organizing maps and neural gas

for advanced classification and clustering

PD Dr. Thomas Villmann
Universität Leipzig

Self-Organizing Maps (SOMs) provide a powerful tool for non-linear data analysis based on prototypes. They have become a widely applied machine learning/neural network approach for visualization, clustering and classification. In the talk, new extensions of the basis algorithm will be proposed to increase the area of applications. In doing so, we will concentrate on diffferent aspects:

1.) Cost function of SOMs, batch learning and non-standard data (dissimilarity data, time series ...)
2.) Fuzzy classification by fuzzy labeled SOM (FLSOM)
3.) Classification task dependent metric adaptation - relevance learning

The approaches will be described in detail, and exemplary real world applications will give impressions of their usability.

Zeit: Freitag, den 25. Mai 2007, 12 Uhr c.t.

Ort: Institut für Neuro- und Bioinformatik
        Seminarraum (1. OG, Raum 17),
        Ratzeburger Allee 160 (Geb. 64, 1. OG)