The Output on the convolutional layer will likely be passed with the ReLU activation operate to bring non-linearity into the model. It will take the element map and replaces all the damaging values with zero. Zero-padding—It allows us to regulate the spatial size of the output volume by padding https://financefeeds.com/apple-aapl-stock-analysis-a-rough-start-to-the-year-worst-since-2008/