Skip to main content
. 2019 Mar 7;13:95. doi: 10.3389/fnins.2019.00095

Figure 1.

Figure 1

The extreme left panel depicts a particular input image from the CIFAR-10 dataset with per pixel mean (over the training set) subtracted that is provided as input to the original ANN. The middle panel represents a particular instance of the Poisson spike train generated from the analog input image. The accumulated events provided to the SNN over 1,000 timesteps is depicted in the extreme right panel. This justifies the fact that the input image is being rate encoded over time for SNN operation.