Michal Cernansky's Homepage
Research

Research topics:

  • Parallel Computing
    • Multicore and manycore computing
    • Stream computing
    • GPGPU - CUDA platform
  • Recurrent neural networks
    • Architectures and training methods
    • Adaptation approaches based on Kalman filtration
    • Approaches based on Markovian architectural bias
    • Echo-state networks
    • Long short-term memory networks
  • Symbolic sequence processing
    • Fixed order Markov models and variable length Markov models
    • Fractal prediction machines and neural prediction machines
    • Connectionist models and symbolic sequence processing
    • Natural Language Processing
  • Biologically plausible computation
    • Neural networks trained according to the BCM (Bienenstock, Cooper, Munro) theory
    • Self-organizing maps

Research projects - principal investigator:

Research projects - coinvestigator:

  • Intelligent embeded systems.
    2008 - 2010, VG 1/0822/08 (VEGA).
  • Theoretical studies and applications of echo state neural networks in artificial intelligence and cognitve science.
    2005 - 2007, APVT-20-002504 (APVV).
  • Echo state neural networks.
    2004 - 2006, VG 1/1047/04 (VEGA).
  • Nonlinear methods in the neural network theory.
    2002 - 2004, VG 1/9046/02 (VEGA).
  • Application of nonlinear methods in the neural network theory.
    1999 - 2001, VG 2/6018/99 (VEGA).