Jan Koutnik is currently a postdoctoral researcher
Juergen Schmidhuber's group at IDSIA. He is focused on research in machine learning, namely artificial neural networks and evolutionary algorithms.
source: home archive
He obtained his Ph.D. computer science in the year 2008 from Faculty of Electrical Engineering, Czech Technical University in Prague,
where he worked as an assistant professor at the Department of Computer Science and Engineering, while being member of Computational Intelligence Group research group.
He czeched out and joined IDSIA in April 2009.
He likes cycling (mountain and road), amateur photography (especially portraits). Check
out his web log, containing tricks for computers and real life.
Faustino Gomez, Jan Koutnik and Juergen Schmidhuber,
Compressed Network Complexity Search,
Proceedings of the 12th International Conference on Parallel Problem Solving from Nature
(PPSN, Taormina, IT), 2012
Julian Togelius, Sergey Karakovskiy, Jan Koutnik and Juergen Schmidhuber,
Super Mario Evolution,
Proceedings ot the IEEE Symposium on Computational Intelligence and Games (CIG), 2009
Publications at CTU in Prague
Pavel Kordik, Jan Koutnik, Jan Drchal, Oleg Kovarik, Miroslav Cepek,
Miroslav Snorek (2010). Meta-learning approach to neural network
optimization. Neural Networks, 23(4), p. 568-582.
Jan Drchal, Ondrej Kapral, Jan Koutnik and Miroslav Snorek
(2009). Combining Multiple Inputs in HyperNEAT Mobile Agent
Controller. In: ICANN '09 Proceedings of the 19th International
Conference on Artificial Neural Networks, vol. 2, p. 775-783,
Springer, Berlin, ISSN 0302-9743
Jan Drchal, Jan Koutnik and Miroslav Snorek
(2009). HyperNEAT Controlled Robots Learn How to Drive on Roads in
SimulatedEnvironment. In: 2009 IEEE Congress on Evolutionary
Computation, p. 6, Research Publishing Services, Singapore, ISBN
Zdenek Buk, Jan Koutnik and Miroslav Snorek (2009). NEAT in
HyperNEAT Substituted with Genetic Programming. In:Adaptive and
Natural Computing Algorithmsvol.5495, nr. , p. 243-252, Springer,
Jan Koutnik and Miroslav Snorek (2008). Temporal Hebbian
Self-Organizing Map for Sequences. In: 16th International
Conference on Artificial Neural Networks Proceedings (ICANN 2008),
Part I, p. 632--641, Springer Berlin / Heidelberg, ISBN
Jan Koutnik and Miroslav Snorek (2007). Extraction of Markov
Chain from Temporal Hebbian Self-organizing Map. In: Proceedings of
the International Workshop on Modelling and Simulation in
Management, Informatics and Control, , 2007. ISBN
Jan Koutnik: Inductive Modelling of Temporal Sequences by
Means of Self-organization (2007). In: Proceeding of Internation
Workshop on Inductive Modelling (IWIM 2007), p. 269-277, CTU in
Prague, ISBN 978-80-01-03881-9
Jan Koutnik and Miroslav Snorek: New Trends in Simulation of
Neural Networks (2007). In: Proceedings of 6th EUROSIM Congress on
Modelling and Simulation ,Ljubljana, ISBN 3-901608-32-X
Jan Drchal, Pavel Kordik and Jan Koutnik
(2007). Visualization of Diversity in Computational Intelligence
Methods. In: Proceedings of 2nd ISGI, International CODATA Symposium
on Generalization of Information, p. 20-34, CODATA Germany, ISBN
Radek Trnka and Jan Koutnik (2006). Application of the
Kohonen's self-organizing map and the group of adaptive models
evolution in social cognition research. Psychologia vol. 4
nr. , p. 238-251, Department of Cognitive Psychology in Education,
Psychologia Society, Kyoto University, Kyoto 606-8501, Japan, ISSN
Jan Koutnik, Roman Mazl and Miroslav Kulich (2006). Building
of 3D Environment Models for Mobile Robotics Using
Parallel Problem Solving from Nature - PPSN-IX.
Heidelberg, p. 721-730, Springer, 2006. ISBN 3-540-38990-3
Jan Koutnik and Miroslav Snorek (2006). Self-Organizing
Neural Networks for Signal Recognition. In: 16th International
Conference on Artificial Neural Networks Proceedings (ICANN
2006) , Part I, p. 406-414, Springer Berlin / Heidelberg,
2006. ISBN 978-3-540-38625-4
Jan Koutnik and Miroslav Snorek (2005) .Neural Network
Generating Hidden Markov Chain. In: Adaptive and Natural Computing
Algorithms - Proceedings of the International Conference in
Coimbra, p. 518-521, Wien: Springer, 2005.
Jan Koutnik and Miroslav Snorek (2004): Efficient Simulation
of Modular Neural Networks. In: Proceedings of the 5th EUROSIM
Congres Modelling and Simulation, Vienna:
EUROSIM-FRANCOSIM-ARGESIM, ISBN 3-901608-28-1
Jan Koutnik and Miroslav Snorek: Single Categorizing and
Learning Module for Temporal Sequences. In: Proceedings of the
International Joint Conference on Neural Networks, p. 2977-2982,
Piscataway: IEEE, 2004. ISBN 0-7803-8360-5
Jiri Kubalik and Jan Koutnik (2003): Automatic Generation of
Fuzzy Rule Based Classifiers by Evolutionary
Algorithms. In: Intelligent and Adaptive Systems in Medicine,
p. 197-206, Praha: CVUT FEL, ISSN 1213-3000
Jan Koutnik and Miroslav Snorek (2003): Enhancement of
Categorizing and Learning Module (CALM) - Embedded Detection of Signal
Change. In: IJCNN 2003 Conference Proceedings, p. 3233-3237,
Piscataway: IEEE, 2003. ISBN 0-7308-7899-7
Jiri Kubalik, Jan Koutnik and Leon J. M. Rothkrantz:
Grammatical Evolution with Bidirectional
Representation. In: Genetic Programming, Proceedings of
EuroGP'2003, p. 354-363, Berlin: Springer, 2003. ISBN
Jan Brunner and Jan Koutnik (2002): SiMoNNe - Simulator of
Modular Neural Networks. In: Neural Network World, vol. 12
nr. 3, p. 267-278, ISSN 1210-0552
Jan Koutnik, Jan Brunner and Miroslav Snorek (2002): The
GOLOKO Neural Network for Vision - Analysis of
Behavior. In: Proceedings of the International Conference on
Computer Vision and Graphics, p. 437-442, Gliwice: Silesian
Technical University, ISBN 83-9176-831-7
HUMANOBS (Humanoids that Learn Socio-Communicative Skills by Observation).
The objective of the HUMANOBS project is to develop a new cognitive architectural principles to allow intelligent agents to learn socio-communicative skills by observing and imitating people.
IDSIA part covers so-called correlator, which is a sequence predictor that extracts rules from input
STIFF (Enhancing biomorphic agility through variable stiffness).
The goal of the proposed study is to equip a highly biomimetic robot hand-arm system with the agility, robustness and versatility that are hallmarks of the human motor system by understanding and mimicking the variable stiffness paradigms that are so effectively employed by the human CNS.
IM-CLeVeR (Intrinsically Motivated Cumulative Learning Versatile Robots)
IM-CLeVeR aims to develop a new methodology for designing robots controllers that can cumulatively learn new efficient skills through autonomous development based on intrinsic motivations, and reuse such skills for accomplishing multiple, complex, and externally-assigned tasks. During skill-acquisition, the robots will behave like children at play which acquire skills autonomously on the basis of "intrinsic motivations". During skill-exploitation, the robots will exhibit fast learning capabilities and a high versatility in solving tasks defined by external users due to their capacity of flexibly re-using, composing and readapting previously acquired skills.
The project investigatesthree fundamental scientific and technological issues:
the mechanisms of abstraction of sensory information,
In state-of-the-art neuroevolution, researchers look for efficient way of encoding
the artificial neural networks in strings (genomes) of symbols (genes) in order to reduce
the search space of such genomes. Jan's recent research in indirect encoding
lead in a method, which describes a neural network weight matrix by a limited set of it's
frequency coefficient. The genome consists of a limited set of frequency coefficients that
transform to the weight matrix using inverse Fourier-type frequency transform.
The weight matrix get decorrelated after transformed to the frequency domain.
The complexity of a genome could be pushed down by encoding the frequency coefficients
with a limited number of bits. If the space of coefficients is small (say less than 32 bits), then it
could be searched exhaustively starting from the shortest bit strings.
Surprisingly, some of the well known benchmarks could be solved with networks described by fairly
short bit-string. For example, single-pole balancing controller consisting of one neuron could be described by just 1 bit (positive constant weights matrix), which means that single-pole benchmark no longer exists.
Temporal Hebbian Self-organizing Map
Temporal Hebbian Self-organizing Map is a recurrent extension of Kohonen's SOM. Additional layer of full recurrent connections
among the nodes is trained in a Hebbian way. The connections accumulate first-order statistics of transitions between states represented by the neurons, while placing the neurons into centroids of clusters using the input connections. The network clusters the data in both input space and time.
The initial THSOM model Hebbian training was improved by Ferro et al. introducing neighborhood in the temporal connections.