In this chapter, we went over three application examples of autoencoder networks. The first type of autoencoder involved a dimension reduction application. Here, we used an autoencoder network architecture that only allowed us to learn about the key features of the input image. The second type of autoencoder was illustrated using MNIST data containing images of numbers. We artificially added noise to the images of numbers and trained the network in such a way that it learned to remove noise from the input image. The third type of autoencoder network involved image correction application. The autoencoder network in this application was trained to remove a black line from input images.

In the next chapter, we will go over another class of deep networks, called transfer learning, and use them for image classification.