In this chapter, we covered transfer learning and demonstrated how using a network pre-trained on a source domain can greatly improve the training time, and ultimately the performance of our deep neural network. I hope you enjoy this technique, it's one of my favorites because it's very practical and I typically get great results from it.
In the next chapter, we will move from computer vision to networks that can remember previous inputs, making them ideal for predicting the next item in a sequence.
