We have almost come to the end of our journey in deep learning with R. This chapter is a bit of a mixed bag of topics. We will begin this chapter by revisiting an image classification task and building a complete image classification solution image files rather than tabular data. We will then move on to explaining transfer learning, where you can use an existing model on a new dataset. Next we discuss an important consideration in any machine learning project - how will your model be used in deployment, that is, production? We will show how to create a REST API that allows any programming language to call a deep learning model in R to predict on new data. We will then move on to briefly discussing two other deep learning topics: Generative Adversarial Networks and reinforcement learning. Finally, we will close this chapter and the book by providing a few other resources that you may be interested in.
In this chapter, we will cover the following topics:
- Building a complete image classification solution
- The ImageNet dataset
- Transfer learning
- Deploying TensorFlow models
- Generative adversarial networks
- Reinforcement learning
- Additional deep learning resources