V O R T R A G
**********************
Oesterreichisches Forschungsinstitut fuer Artificial Intelligence(OFAI) der OSGK
Freyung 6/6, A-1010 Wien
Tel: +43-1-5336112-17, Fax: +43-1-5336112-77, Email: sec@ofai.at
Nenad Tomasev
Artificial Intelligence Laboratory
Jozef Stefan Institute
Ljubljana, Slovenia
"THE ROLE OF HUBS IN HIGH-DIMENSIONAL DATA ANALYSIS"
Many important data types are inherently high-dimensional, as they
require many features to be properly represented. This includes, but
is not limited to: text, images, sensor data, music/audio files, etc.
Analyzing such high-dimensional data is known to be difficult, for
many reasons. Sparsity might be the most obvious problem, though there
are other, more subtle issues at hand. Inherently high-dimensional
data gives rise to hubs, very influential points in the k-nearest
neighbor topology. Hub-points are retrieved very often during
similarity search, while most other points receive little to no
attention during the learning process. This skewed distribution of
influence affects many data mining methods, often in very negative
ways. We will show some initial steps that we have been taking
towards designing hubness-aware data mining and machine learning
methods. This includes classification, clustering and metric learning.
*********
Time: Thursday, 28th February 2013, 6.30 p.m. sharp
Location: Oesterreichisches Forschungsinstitut
fuer Artificial Intelligence, OFAI
Freyung 6, Stiege 6, 1010 Wien
OESTERREICHISCHES FORSCHUNGSINSTITUT
FUER ARTIFICIAL INTELLIGENCE
Univ.-Prof. Ing. Dr. Robert Trappl, MBA
*********