Information, Complexity and Learning

Group Leader: Teemu Roos

The Information, Complexity and Learning (ICL) research group is a part of the Cosco research group and studies the theory and applications of probabilistic models, especially graphical models. A particular area of interest is information theoretic methods. Application domains where we have continued collaboration include human–computer interaction, digital humanities, and bioinformatics.

News and Events (see also Cosco News on the right)

  • October 26, 2018: The Elements of AI reaches 100k participants!
  • August, 2018: Elias Jääsaari moves to Kvasir Ltd in Cambridge. We wish Elias good luck!
  • May 15, 2018: The Elements of AI online course is launched in collaboration with Reaktor.
  • April 2018: Elias Jääsaari presents "Minimax optimal Bayes mixtures for memoryless sources over large alphabets" at ALT-2018, Lanzarote.
  • March 26, 2018: Teemu Roos gives an invited talk at Google Inc. (Pittsburgh)
  • March 21, 2018: Teemu Roos gives an invited talk at Apple Inc. (Apple Park)
  • Papers accepted to Digital Scholarship in the Humanities ("Quantitative methods for the analysis of Medieval calendars") and the forthcoming book Online Distribution of Content in the EU ("Digital distribution of AI-generated content: Authorship and inventorship in the age of artificial intelligence")
  • Ville Hyvönen visits Sanjoy Dasgupta's group at UC San Diego in February-April 2018
  • Papers accepted to ALT 2018 ("Minimax optimal Bayes mixtures for memoryless sources over large alphabets") and AISTATS 2018 ("Quotient normalized maximum likelihood criterion for learning Bayesian network structures")
  • December 13, 2017: Teemu Roos gives a keynote talk "Forget about the Terminator already! AI Education for All" at the 1st AI Day Finland. (Slides on SlideShare)
  • November 8, 2017: Teemu Roos discusses the past, present, and future of AI in an public event at @ThinkCorner (in Finnish)
  • 10/2017: The Academy of Finland awards 2-year funding for out joint proposal with Prof Fluyra Djurabekova and Prof Kai Nordlund (Department of Physics) on "Machine Learning for Quantum Mechanics Based Material Design"
  • 6/2017: The Academy of Finland awards 4-year funding for our joint proposal with Prof. Tapio Pahikkala (Turku) on "Tensor-Based Machine Learning for Big Data with Inherent Dependencies".
  • Papers accepted to International Journal of Approximate Reasoning, New Generation Computing, and Pattern Recognition Letters (see below).
  • 6/2017: Jukka Kohonen joins the ICL group as a postdoc. Welcome Jukka!
  • We will co-organize the 10th WITMSE workshop in Paris on September 11-13, 2017.
  • 4/2017: Our joint proposal with Prof. Pulkit Grover (CMU) on ''Efficient and Robust Cognitive IoT Systems using Unreliable Sensors'' is funded by the Academy of Finland and NSF.
  • Teemu Pitkänen moves to Google Research Europe in Zürich. We wish Teemu good luck!
  • 3/2017: Teemu Roos teaches at the 21st CIMO Winter School, Tvärminne Zoological Station, Finland
  • March 3rd, 2017: Yuan Zou defended her doctoral thesis on "Model Selection for Bayesian Networks and Sparse Logistic Regression". Congratulations to Yuan!
  • 2/2017: Teemu Roos gives a talk at the Workshop on Mathematical Approaches to Evolutionary Trees and Networks in Banff, Canada. Video available.
  • 10/2016: Joonas Miettinen joins the group as a PhD student supervised by Teemu Roos and Janne Pitkäniemi (Finnish Cancer Registry). Welcome Joonas!
  • Former ICL postdoc Sotiris Tasoulis is a co-chair of the 2nd Workshop on Advances in High Dimensional Big Data on December 5-8, 2016.
  • Teemu Roos gives talks at the University of Melbourne (details) on October 6, 2016 and Monash University on October 12, 2016. 
  • "Minimum Description Length Principle" by Teemu Roos appears in the Encyclopedia of Machine Learning and Data Mining.
  • We are organizing the 9th Workshop on Information Theoretic Methods in Science and Engineering (WITMSE-2016) in Helsinki on September 19-21, 2016.
  • May 27th 2016: Jussi Määttä will defend his PhD thesis "Model selection methods for linear regression and phylogenetic reconstruction" (2pm, Exactum B123).
  • May 17th 2016: Teemu Roos gives a talk on "Directed acyclic graphs as a model for biological and cultural evolution" at Purdue University, USA.
  • April 24th 2016: Teemu Roos gives a talk on "Machine learning and the evolution of fairy tales" in the Informatics Colloquium at the Martin Luther University, Halle, Germany.
  • Summer 2016 internship to Elias Jääsaari. Welcome Elias!
  • Papers accepted to PLOS ONE, IEEE Transactions on Big Data, and EuroVis 2016 conference.
  • Ville Hyvönen joins the group as a PhD student under the Scalable Probabilistic Analysis project. Welcome Ville!

(Click here for older events.)


Senior Members








Teemu Roos,
Associate professor

Jukka Kohonen, PhD




Jussi Määttä, PhD



Ville Hyvönen
PhD student

Janne Leppä-aho
PhD student

Joonas Miettinen
PhD student




Past members & alumni

Elias Jääsaari (MSc 2018) >> Kvasir Ltd (Cambridge)
Yuan Zou (PhD 2017)
Teemu Pitkänen (MSc 2016) >> Google Research Europe (Zurich)
Jussi Määttä (PhD 2016) >> Data Scientist @
Santeri Räisänen (research assistant 6/2014–12/2015)
Yang Zhao (research assistant 6/2014–12/2015)
Sotiris Tasoulis (postdoc 9/2013–8/2015) >> Liverpool John Moores University
Quan (Eric) Nguyen (MSc, research assistant 5/2013–8/2015) >> TU Eindhoven
Arttu Modig (MSc, research assistant 1/2012–5/2015) >> Finland's Slot Machine Association
Simo Linkola (MSc, research assistant 1/2011-4/2014) >> Computational Creativity Group
Nikolay Vasilev (intern summer 2013)


Recent Papers and Preprints


E. Jääsaari, V. Hyvönen, T. Roos (2019). Efficient autotuning of hyperparameters in approximate nearest neighbor search, to appear in Proc. 23rd Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD-2019), Macau, China.  C++ library with Python bindings

T. Heikkilä and T. Roos (2018), Quantitative methods for the analysis of Medieval calendars, Digital Scholarship in the Humanities, 33(4):766–787.

U. Sheth, S. Dutta, M. Chaudhari, H. Jeong, Y. Yang, J. Kohonen, T. Roos, and P. Grover (2018). An application of storage-optimal MatDot codes for coded matrix multiplication: Fast k-nearest neighbors estimation, to appear in IEEE Big Data Conference, Seattle, December 10–13, 2018

J. Leppä-aho, S. Räisänen, X. Yang, and T. Roos (2018). Learning non-parametric Markov networks with mutual information, in Proc. 9th Int Conf on Probabilistic Graphical Models, Prague, September 11–14, 2018. Best student paper award

E. Jääsaari, J. Leppä-aho, T. Silander and T. Roos (2018). Minimax optimal Bayes mixtures for memoryless sources over large alphabets, in Proc. Int. Conf. on Algorithmic Learning Theory (ALT 2018)
T. Silander, J. Leppä-aho, E. Jääsaari, and T. Roos (2018). Quotient normalized maximum likelihood criterion for learning Bayesian network structures, in Proc. 21st Int. Conf. on Artificial Intelligence and Statistics, PMLR 84:948-957 (AISTATS 2018) (+ supplementary PDF

J. Leppä-aho, S. Räisänen, X. Yang, and T. Roos (in preparation). Learning non-parametric Markov networks with mutual information, arXiv:1708.02497.

Y. Zou, J. Pensar, and T. Roos (2017). Representing local structure in Bayesian networks by Boolean functions, accepted to Pattern Recognition Letters.

J. Leppä-aho, J. Pensar, T. Roos, and J. Corander (2017). Learning Gaussian graphical models with fractional marginal pseudo-likelihoodInternational Journal of Approximate ReasoningarXiv:1602.07863

Y. Zou and T. Roos (2017). On model selection, Bayesian networks, and the Fisher information integral, New Generation Computing 35(1) (Special Issue on AMBN 2015), January 2017.


V. Hyvönen, T. Pitkänen, S. Tasoulis, E. Jääsaari, R. Tuomainen, L. Wang, J. Corander, and T. Roos. Fast nearest neighbor search through sparse random projections and voting, 2016 IEEE International Conference on Big Data (IEEE Big-Data 2016), Washington DC, Dec. 5–8.  C++ library (with Python bindings) | benchmarks

T. Roos (2016). Minimum Description Length Principle, in Sammut, C. and Webb, G.I. (eds), Encyclopedia of Machine Learning and Data Mining.

T. Heikkilä and T. Roos, (2016). Thematic Section on Studia Stemmatologica, Digital Scholarship in the Humanities 31(3):520–522, doi:10.1093/llc/fqw038.

L. Wang, S. Tasoulis, T. Roos, and J. Kangasharju (2016). Kvasir: Scalable provision of semantically relevant web content on big data framework, IEEE Transactions on Big Data.

Y. Zhao, S. Tasoulis, and T. Roos (2016). Manifold visualization via short walks, EuroVis-2016.

J. Määttä and T. Roos (2016). Maximum parsimony and the skewness test: A simulation study of the limits of applicability, PLOS ONE 11(4):e0152656.

Y. Zou and T. Roos (2016). Sparse Logistic Regression with Logical Features, Proc. 20th Pacific Asia Conference on Knowledge Discovery and Data Mining (PAKDD-2016).

J. Määttä and T. Roos (2016). Robust Sequential Prediction in Linear Regression with Student’s t-distribution, in Proc. 14th International Symposium on Artificial Intelligence and Mathematics (ISAIM 2016).


R. Eggeling, T. Roos, P. Myllymäki, and I. Grosse (2015). Inferring intra-motif dependencies of DNA binding sites from ChIP-seq data, BMC Bioinformatics.

Y. Zou and T. Roos (2015). On model selection, Bayesian networks, and the Fisher information integral, in Proc. 2nd Workshop on Advanced Methodologies for Bayesian Networks (AMBN-2015).

J. Määttä, D. F. Schmidt, and T. Roos, (2015). Subset Selection in Linear Regression using Sequentially Normalized Least Squares: Asymptotic Theory, Scandinavian Journal of Statistics.

J. Tehrani, Q. Nguyen, and T. Roos, (2015). Oral fairy tale or literary fake? Investigating the origins of Little Red Riding Hood using phylogenetic network analysis, Digital Scholarship in the Humanities.

Q. Nguyen and T. Roos, (2015). Likelihood-based inference of phylogenetic networks from sequence data by PhyloDAG, in Proc. 2nd International Conference on Algorithms for Computational Biology (AlCoB-2015).

K. Watanabe abd T. Roos, (2015). Achievability of asymptotic minimax regret by horizon-dependent and horizon-independent strategiesJMLR.

(Click here for older papers.)

Last updated on 21 Dec 2018 by Teemu Roos - Page created on 6 Sep 2012 by Petri Myllymäki