CNN’s Glossary

CNN’s Glossary CNNs: Convolutional neural network. That is, a network which has at least one convolutional layer. A typical CNN also includes other types of layers, such as pooling layers and dense layers. Convolution: The process of applying a kernel (filter) to an image Kernel / filter: A matrix which is smaller than the input, used to transform…

Defining a CNN Network architecture

Convolutional Neural networks will make use of the following types of layers: Convolutional Layers Maxpooling layers Fully connected (linear) layers To define a network in PyTorch you create a class: class Net(nn.Module) Define the layers on the init function def __init__(self, n_classes): super(Net,self).__init__() # 1 input channel (grayscale image), 32 outputs or feature maps, 5×5 conv. kernel…