Colloquium Computer Science, Bassam Mokbel and Xibin Zhu, Bielefeld University, Germany

Date:                           Monday, June 24^th 2013
Speaker:                     Bassam Mokbel and Xibin Zhu, Cognitive Interaction Technology Center of Excellence (CITEC), Bielefeld University, Bielefeld, Germany
Room:                         5161.0267 (Bernoulliborg)
Time:                          16.00

Title: Dimensionality reduction to visualize large data sets

Abstract:

Due to rapidly developing sensor and storage technologies, digital data is available

in large quantities in virtually all academic and industrial disciplines. In order to make

these large (and typically high-dimensional) data sets easily accessible for the respective

field experts, a very active branch of data analysis research is concerned with embed-

ding high-dimensional points as two- or three-dimensional vectors for visualization.

This way, the data set can be displayed as a simple scatter plot, which the user can

inspect visually. Based on a reliable visual representation, human experts can reason

about certain characteristics of the original data, for example by revealing large-scale

structures or investigating close neighborhood relations.

In our talk, we give a brief overview of some well-established dimensionality re-

duction techniques and show examples for real-world applications. We address the

ill-posedness of the task and discuss formal evaluation criteria to assess the reliability

of the embedding. Since many dimensionality reduction methods are based on pairwise

proximities between the data, we will introduce approximation techniques to tackle the

problem of the inherent quadratic complexity of these algorithms.

Colloquium coordinators are Prof.dr. M. Aiello (e-mail : M.Aiello rug.nl ) and
Prof.dr. M. Biehl (e-mail: M.Biehl rug.nl )

http://www.rug.nl/research/jbi/news/colloquia/computerscience