University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Colloquium of the Department and the PhD Program

title

Density-based clustering and its application to high-dimensional data analysis

speaker

Dr. Peer Kröger, LMU München
München, Germany

date & place

Wednesday, 07.12.2005, 16:15 h
Room C252

abstract

Clustering is one of the major data mining tasks and aims at grouping the data objects into meaningful classes (clusters) such that the similarity of objects within clusters is maximized, and the similarity of objects from different clusters is minimized. While most traditional clustering methods usually work fine in relatively low dimensional feature spaces, they tend to have problems with high dimensional data sets. In particular, most traditional methods fail to detect meaningful clusters in high dimensional data sets due to the curse of dimensionality. Since many today's real-world data sets are high dimensional novel clustering solutions are required that can handle these problems of high dimensionality. Thus, in recent years, the task of subspace clustering or projected clustering have been explored, where clusters are detected in arbitrary projections of the entire feature space.

In this talk, the basic notion of density-based clustering which has turned out to be one of the most successful traditional approaches to clustering is reviewed. It is then discussed, how this density-based notion can be extended to detect (subspace/projected) clusters in high dimensional spaces. In addition, this talk provides the definition of a relatively new area of research, namely the task of correlation clustering. Correlation clusters are sets of points that fit to a common hyperplane of arbitrary dimensionality. It will also be shown, how density-based clustering can be extended for correlation clustering.