Appendix

Go to a chapter.

Every chapter in the book. Pick a chapter or jump straight to one of its modules.

• Chapter 1 · coming soonWhy not a dense layer?Flatten an image and watch the parameter count explode.
• Chapter 2 · coming soonConvolution by handSlide a kernel and compute each output as a dot product.
• Chapter 3Stride, padding and dilationThree settings that decide where the kernel lands and how big the output is.

  1. 3.1Stride
  2. 3.2Padding
  3. 3.3Dilation
  4. 3.4Putting it together
• Chapter 4 · coming soonChannelsWhy kernels are really three-dimensional.
• Chapter 5PoolingMax and average pooling, and which input wins.

  1. 5.1Max pooling
  2. 5.2Average pooling
  3. 5.3Small shifts
  4. 5.4Pool or stride?
• Chapter 6Receptive fieldsStack convolutions and pooling layers, and watch how much of the image one output can see.

  1. 6.1One layer
  2. 6.2Stacking layers
  3. 6.3Stride and pooling
  4. 6.4Not every pixel counts
• Chapter 7 · coming soonBackprop through a convolutionWhere the gradients go.
• Chapter 8 · coming soonUpsampling and transposed convolutionSpreading values back out, and checkerboard artifacts.
• Chapter 9 · coming soonWhat the network learnedA real CNN running in the browser, layer by layer.
• Chapter 10 · coming soonArchitectures as storiesLeNet to ResNet to U-Net, one idea at a time.