Left panel: building block of a fully-connected neural network. Each input of the previous layer is connected to each neuron of the output. Each connection is represent by different lines where the width is associated to higher weights and the dashed lines to negative weights. Right panel: three-dimensional convolution carried out by a convolutional layer. The 3D-kernel traverses the whole input, producing a single scalar at each position. At the end, a 2D feature map will be created for each 3D kernel. When all feature maps are stacked, a feature map tensor will be created.