Brief description of STDP

        Different from the traditional gradient descent method, the spike neural network usually uses a more biological STDP (spike timing dependent plasticity) learning strategy to update the connection weights of the network.

          The most commonly used method for updating in spiking neural networks is the STDP method. The way STDP updates the synaptic weights is: if the pre-synaptic pulse arrives earlier than the post-synaptic pulse, it will lead to Long-Term Potentiation (LTP) effect, that is  The weight will increase. Conversely, if the presynaptic pulse arrives later than the postsynaptic pulse, it will cause LTD, that is,  the weight will decrease.

        The most basic STDP is:

        where,  is the post-synaptic pulse firing time,  is the pre-synaptic pulse firing time, and W(x) is:

        again:

        which ends up being:

 In this way, the relationship between the weight change and the front and rear pulses is established, so that the SNN weight update can be calculated by the above formula.

 Reference path: Spike-timing dependent plasticity - Scholarpedia