The From Data to Knowledge FDK Research Unit a National Center of Excellence

1
The From Data to Knowledge (FDK) Research Unit
a National Center of Excellence

• Esko Ukkonen

Department of Computer Science
2
Structure of the FDK

• national Center-of-Excellence status (Academy of
  Finland) for 2002-2007 basic funding 267 k /
  year
• host institutions
• University of Helsinki, Dept of Computer Science
• Helsinki University of Technology, Laboratory of
  Computer and Information Science
• about 60 members
• professors
• Esko Ukkonen (director, academy professor -2004)
  
• Heikki Mannila (academy professor 2004 -)
• Hannu Toivonen
• Helena Ahonen-Myka
• Juho Rousu (2005 -)
• Tapio Elomaa -gt Tampere Univ of Technology
• Jaakko Hollmén (HUT)

3
Mission and goals

• The FDK unit develops methods for forming useful
  knowledge from large masses of data. The unit
  operates in multi-disciplinary fashion,
  integrating in its research groups excellence in
  computational methods, statistical techniques,
  and application sciences.
• data gt computational methods gt knowledge
• problem gt concepts and formalization gt
  algorithm gt algorithm analysis gt
  implementation gt evaluation in practice
• Bioinformatics PhD Programme (1997),
  international Masters Programme (2006)

4
Core competence

• Combinatorial Pattern Matching matching and
  finding patterns in strings and in more complex
  discrete structures, deriving their combinatorial
  properties, and exploiting these to achieve
  superior performance for the corresponding
  computational problems (Esko Ukkonen 1980 - )
• Data Mining finding interesting and useful
  patterns from masses of data (Heikki Mannila 1992
  - )
• gt Combinatorial algorithms probabilistic
  models
• Strong international status Mannila Toivonen
  Ukkonen among top-ten of most cited Finnish
  computer scientists

5
Highlights 25 PhD dissertations from FDK

• Mannila group
• Pirjo Moen Similarity notions for data mining.
• lecturer at CS/UH
• Barbara Heikkinen Document structures and
  document assembly.
• Nokia Research
• Vesa Ollikainen Simulation techniques for
  disease gene localization.
• Center for Scientific Computing
• Marko Salmenkivi Computational methods for
  intensity models.
• postdoc at CS/UH
• Mikko Koivisto Algorithms for the analysis of
  genetic risks.
• Academy postdoc at HIIT/BRU
• HMM techniques for genome analysis

6
Highlights 25 PhD dissertations (cont.)

• Mannila group (cont)
• Jouni Seppänen (HUT) Using and extending
  itemsets in data mining query approximation,
  dense itemsets, and tiles
• postdoc HUT
• Taneli Mielikäinen Summarization techniques for
  pattern collections in data mining
• Nokia Research, Palo Alto, Ca
• Antti Leino On toponymic constructions as an
  alternative to naming patterns in describing
  Finnish lake names
• Center for the Languages of Finland
• Evimaria Terzi Problems and algorithms for
  sequence segmentations
• IBM Almaden

7
Highlights 25 PhD dissertations (cont)

• Toivonen group
• Kari Vasko Computational methods for
  paleoecology.
• Center for Scientific Computing
• private company Ekahau
• Petteri Sevon Association-based gene mapping.
• Karolinska Institutet
• project manager at HIIT/BRU
• Mika Raento Exploring privacy for ubiquitous
  computing tools, methods and experiments
• private company Jaiku

8
Highlights 25 PhD dissertations (cont)

• Elomaa group
• Juho Rousu Range partitioning in classification
  learning.
• VTT Biotech, postdoc London/Southampton (Marie
  Curie)
• now professor of bioinformatics at CS/UH
• dataflow techniques for metabolic modeling
• machine learning for structured data
• Matti Kääriäinen Learning small trees and graphs
  that generalize.
• postdoc at International Computer Science
  Institute (ICSI) of UC Berkeley (Richard Karps
  group)
• postdoc at HIIT

9
Highlights 25 PhD dissertations (cont)

• Ahonen-Myka group
• Antonin Doucet Advanced document description, a
  sequential approach
• postdoc at INRIA, France
• Miro Lehtonen Indexing heterogeneous XML for
  full-text search
• postdoc at FDK
• Lili Aunimo Methods for answer extraction in
  textual question answering
• research coordinator at EVTEK Polytechnic

10
Highlights 25 PhD dissertations (cont)

• Ukkonen group
• Kimmo Fredriksson Rotation invariant matching.
• Academy postdoc at Univ Joensuu, Finland
• Jaak Vilo Pattern discovery from biosequences.
• European Bioinformatics Institute (UK)
• Egeen Ltd Univ Tartu, Estonia
• Kjell Lemström String matching for music
  retrieval.
• City Univ London
• Academy Research Fellow at CS/UH
• music retrieval
• Veli Mäkinen Parametrized approximate string
  matching.
• postdoc in Bielefeld
• lecturer, postdoc and Academy Research Fellow at
  CS/UH

11
Highlights 25 PhD dissertations (cont)

• Ukkonen group (cont)
• Janne Ravantti Reconstruction of macromolecular
  complexes from electron microscopy images.
• postdoc at Structural biology CoE at Dept Biology
  of UH
• Teemu Kivioja Computational tools for a
  transcriptional profiling method.
• researcher at VTT Biotech and Biomedicum UH
• Hellis Tamm Minimality of multitape finite
  automata.
• postdoc in Tallinn, Estonia
• Ari Rantanen Algorithms for 13C metabolic flux
  analysis
• postdoc ETH, Zurich

12
Future Algorithmic Data Analysis (ALGODAN) new
CoE for 2008-2013

• Sequence analysis
• Learning from and mining structured and
  heterogeneous data
• Discovery of hidden structure in
  high-dimensional data
• Foundations of algorithmic data analysis

cgccgagtgacagagacgctaatcaggctgtgttctcaggatgcgtac