Deep Learning with PyTorch Quick Start Guide

Multiple kernels

In each convolution, we can include multiple kernels. Each kernel in a convolution generates its own feature map. The number of kernels is the number of output channels, which is also the number of feature maps generated by the convolutional layer. We can generate further feature maps by using another kernel. As an exercise, calculate the feature map that would be generated by the following kernel:

By stacking kernels, or filters, and using kennels of different sizes and values, we can extract a variety of features from an image.

Also, remember that each kernel is not restricted to one input dimension. For example, if we are ...