The output of the convolutional layer is often handed with the ReLU activation function to bring non-linearity for the model. It requires the attribute map and replaces many of the adverse values with zero. Zero-padding—It allows us to regulate the spatial size in the output volume by padding zeros https://financefeeds.com/priced-at-just-0-175-this-coin-is-gaining-ground-to-beat-shiba-inu-shib-and-turn-670-into-13400-in-7-weeks/