Συντάχθηκε 04-02-2013 18:22
από Emmanouil Varouchakis
Email συντάκτη: evarouchakis<στο>tuc.gr
Ενημερώθηκε:
-
Κύρια: ΕΔΙΠ ΜΗΠΕΡ.
Άλλες ιδιότητες: απόφοιτος ΜΔΕ/Διδ. ΜΗΧΟΠ
Ανάρτηση εκ μέρους του Καθηγητή Διονυσίου Χριστόπουλου.
POSITION IN Algorithm design and Development in
*Approximate Nearest Neighbor Search and Adaptive Data structures*
Duration: Up to 9 months, available now.
Funding: "Computational Geometry Learning" (http://cglearning.eu/)
Lab:
http://erga.di.uoa.gr
Adviser. Ioannis Emiris (http://www.di.uoa.gr/~emiris/index-eng.html)
TOPIC
We study Approximate nearest neighbors beyond 3 dimensions, where we
exploit structural information, namely coherence (non-randomness) of
the data points. This is an important problem with applications in
many areas, including optimization and searching, machine learning,
and geometric inference. Data structures and query algorithms will be
developed, with the ultimate goal of implementing them in the C++ CGAL
Library, starting with an experimental analysis of algorithms for data
of low intrinsic dimension, e.g. points distributed on (unknown)
lines. We aim at obtaining good performance in practice in key
application areas. Since there are many theoretical and implementation
issues, the project may be adapted to the profile of the candidate.
--
Ioannis Emiris
http://www.di.uoa.gr/~emiris/index-eng.html