Spike-timing dependent plasticity with release probability supported to eliminate weight boundaries and to balance the excitation of Hebbian neurons

Spike-timing-dependent plasticity is considered as the key underline mechanism which processes the signals in brain. With the introduction of spike-timing dependent plasticity as a long-lasting synaptic modification, neural networks have been driven to era of processing information on the basis of relative timing between presynaptic and postsynaptic action potentials. One of the main drawbacks that impinged the successive progress of the researches in this area is the constraints that have been put on the weight algorithms of these networks. Here, we analyzed the possibility of eliminating these constraints from the neural networks by introducing release probability and dynamic multiple stochastic synaptic connections between neurons. Our results have proven the possibility of balancing the excitation of the neural networks as our modeled network stabilizes its weights distribution for Poisson inputs with frequency less than 40 Hz. Further, the excited synapses have resembled the median of the weight distribution into unimodal Gaussian distribution for input frequencies between 15 Hz to 40 Hz.