Spike-Based Approximate Backpropagation Algorithm of Brain-Inspired Deep SNN for Sonar Target Classification

Figure 4

A simple example of the decoding process. In each time step, the value of the output layer is the result of multiplying the input spike train of the preneuron by the synaptic weight of the output layer. In the decoder, the values of the output layer in all time steps are accumulated and divided by the total time steps. Finally, the classification results are predicted by comparing the values of neurons in the decoder.