In Part 2 of Embedded Machine Learning, I speak about one important technique to prune a neural network and perform inference on small devices.