Understanding neural networks through visualization

  • Objective: Understand neural mappings, error surfaces, data structures through visualization.
  • Participants: W. Duch and co-authors
  • Time: started in 2003, some work done earlier, on-going project
  • Results:

    VISER Toolbox for visualization of time series data.

    Neurodynamics, time-dependent:

    1. Dobosz K, Duch W, Visualization for Understanding of Neurodynamical Systems.
      Cognitive Neurodynamics 5(2), 145-160, 2011.
    2. Duch W, Neurodynamics and the mind.
      Proc. of the International Joint Conference on Neural Networks, San Jose, CA, IEEE Press, pp. 3227--3234, 2011.
      Presented at the IJCNN 2011, San Jose, special session "What Neural Modeling Tells Us about Ourselves''.
    3. Duch W, Dobosz K, Attractors in Neurodynamical Systems.
      Advances in Cognitive Neurodynamics II (eds. R. Wang, F. Gu), pp. 157-161, 2011
    4. Dobosz K, Duch W. (2010) Understanding Neurodynamical Systems via Fuzzy Symbolic Dynamics.
      Neural Networks Vol. 23 (2010) 487-496, 2010
      http://dx.doi.org/10.1016/j.neunet.2009.12.005
    5. Duch W, Dobosz K, Jovanovic A, Klonowski W. (2010) Exploring the landscape of brain states
      NeuroMath COST Action BM0601, archived in Frontiers in Neuroscience.

    6. Dobosz K, Duch W, Fuzzy Symbolic Dynamics for Global Visualization.
      Lecture Notes in Computer Science, vol. 5164, 471-478, 2008.
      Presented at: International Conference on Artificial Neural Networks (ICANN'08), Prague, Czech Republic
    7. Dobosz K, Duch W, Global Visualization of Neural Dynamics.
      Neuromath Workshop, Dornburg Castle, Jena, Germany, 28-29 April 2008, pp. 15-16

    Feedforward networks.

    1. Duch, W, Coloring black boxes: visualization of neural network decisions.
      International Joint Conference on Neural Networks, Portland, Oregon, 2003, IEEE Press, Vol. I, pp. 1735-1740
      Matlab files with software/data used in this paper, 1 MB. Please note that it requires the Netlab neural network toolbox to run.

    2. Duch W, Visualization of hidden node activity in neural networks: I. Visualization methods.
      The 7th International Conference on Artificial Intelligence and Soft Computing (ICAISC), Zakopane, Poland, June 2004. Eds. L. Rutkowski, J. Siekemann, R. Tadeusiewicz, L. Zadeh. Lecture Notes in AI Vol. 3070 (2004) 38-43
    3. Duch W, Visualization of hidden node activity in neural networks: II. Application to RBF networks.
      The 7th International Conference on Artificial Intelligence and Soft Computing (ICAISC), Zakopane, Poland, June 2004. Eds. L. Rutkowski, J. Siekemann, R. Tadeusiewicz, L. Zadeh. Lecture Notes in AI Vol. 3070 (2004) 44-49
    4. Kordos M, Duch W, A Survey of Factors Influencing MLP Error Surface.
      Control and Cybernetics 33(4) (2004) 611-631.
    5. Kordos M, Duch W, On Some Factors Influencing MLP Error Surface.
      The 7th International Conference on Artificial Intelligence and Soft Computing (ICAISC), Zakopane, Poland, June 2004. Eds. L. Rutkowski, J. Siekemann, R. Tadeusiewicz, L. Zadeh. Lecture Notes in AI Vol. 3070 (2004) 217-222
    6. Duch W, Internal representations of multi-layered perceptrons.
      In: Issues in Intelligent Systems: Paradigms. Eds. O. Hryniewicz, J. Kacprzyk, J. Koronacki, S.T. Wierzchoń, Exit, Warsaw, Poland (2005), pp. 49-62.
    7. F. Piękniewski and L. Rybicki: Visual comparison of performance for different activation functions in MLP networks, IJCNN 2004.

    8. F. Piękniewski and L. Rybicki: Visualizing and Analyzing Multidimensional Output from MLP Networks via Barycentric Projections, ICAISC 2004.

    Also connected with visualization:


    Links to other Duch-Lab projects, many talks and to other papers on various subjects.

  • Working log (local accsess only), maintained by Wlodzislaw Duch.