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

<div>The average number of spike activations at each layer of ResNet11. When using 100 time steps, 1000 samples are randomly selected from the test samples of the CIFAR-10 dataset, and the accuracy of this method is 92.00%; the average number of spikes activated per layer under the ResNet11 network architecture is obtained.</div>

Computational Intelligence and Neuroscience

fig8

Figure 8

Figure 8: Spike-Based Approximate Backpropagation Algorithm of Brain-Inspired Deep SNN for Sonar Target Classification 