Learning Dynamics

o Dynamic properties of neural networks with adapting synapses

    Studied the dynamics of learning or developing  in  a neural network
    when two kinds of dynamic processes take place  ---  the change with
    time of the activity of each neuron and  the  change  in strength of
    the connections (synapses)  between  neurons.   Developed a Lyapunov
    function  to  help  understand  the  combined  activity and  synapse
    dynamics  for  a  class  of  such  adaptive networks,  particularly,
    feedback networks.  Illustrated the methods  and  viewpoint by using
    them  to  describe  the   development   of   columnar  structure  of
    orientation selective cells in  primary  visual  cortex.  Proposed a
    biologically  plausible  model  in  which   the  columnar  structure
    originates from symmetry  breaking  in  feedback  pathways within an
    area of cortex, rather than feedforward pathways between areas.

  ^{Dong D W 1991}
    {Dynamic properties of neural network with adapting synapses}
    {Proc International Joint Conference on Neural Networks, Seattle}
    { Vol~2 pp~255-260}

  ^{Dong D W and Hopfield J J 1992}
    {Dynamics of interconnection development within visual cortex}
    {Proc International Joint Conference on Neural Networks, Baltimore}
    { Vol~3 pp~85-90}

  ^{Dong D W and Hopfield J J 1992}
    {Dynamic properties of neural networks with adapting synapses}
    {Network: Computation in Neural Systems}{ Vol 3(3) pp 267--283}

o Associative decorrelation dynamics


    Proposed a local and  associative  learning  rule,  in networks with
    feedback  connections,  which  dynamically  decorrelates the network
    inputs.   Applied  this  dynamics  to  model  both   the  long  term
    development and the short term  adaptation  of orientation selective
    cells in visual cortex.  Examined the implication  of the hypothesis
    that   the   intracortical   connections   dynamically   decorrelate
    activities  of  orientation  selective  cells.    Showed  that  this
    decorrelation  dynamics   leads   to   quantitative  predictions  of
    orientation contrast and orientation  adaptation  which  are in good
    agreement   with   various   psychophysical  and  neurophysiological
    experiments.

  ^{Dong D W 1992}
    {Hebbian learning in feedback networks: development within visual cortex}
    {Computation and Neural Systems
      Eeckman F H and Bower J M (Eds), Kluwer, Boston, MA}{ pp~383-388}

  ^{Dong D W 1993}
    {Anti-Hebbian dynamics and total recall of associative memory}
    {Proc World Congress on Neural Networks, Portland}{ Vol~2 pp~275-279}

  ^{Dong D W 1994}
    {Associative decorrelation dynamics: a theory of
      self-organization and optimization in feedback networks}
    {In: Advances in Neural Information Processing Systems 7
      (Tesauro G, Touretzky DS, Leen TK, eds) MIT Press, Cambridge, MA}
    {pp~925-932}

( Papers' Index of Dawei Dong )


Send comments to Dawei Dong: dawei@ccs.fau.edu