Article Notes: Ghosh Dastidar, Adeli (2009)

• Multiple spiking neural network
• Each neuron is fully connected and each connection involves multiple synapses
• Each transmission along each synapse involves multiples spikes, unlike in a single spiking SNN which only involves the transmission of one spike.
• Variable meanings:
  • x = internal state
  • j = postsynaptic neuron (receiving transmission)
  • t = time (Maximum computation and efficiency found with time step of 1 ms)
  • N = neuron
  • l = layer
  • i = presynaptic neuron (sending transmission)
  • K = number of synapses, constant for any two neurons
  • k = synapse number
  • G = number of spikes
  • g = spike number
  • w = synapse weight
  • ε = spike response function
  • d = delay
  • And 𝜏 = time decay constant (determines spread shape of the function. These researchers found that the encoding interval + 1 ms worked well)
  • θ = Neuron Threshold
  • η = learning rate (Maximum computation and efficiency found from 0.001-0.01)
• Neuron in layer l are postsynaptic to all presynaptic neurons in l+1 (layers are numbered backward starting with output layer as number 1)
• Modeling of synapses is identical for all neurons, and the kth synapse between any new neurons has the same delay.