Summary

In this chapter, we mainly discussed a new method of distributed machine learning, called federated learning. The key concept of federated learning is that it enables collaborative model training without sharing each worker's local data. Thus, federated learning makes it possible for data privacy applications such as multiple banks to collaboratively train a model for fraud detection, for example.

After reading this chapter, you should understand how federated learning works via sharing knowledge without sharing real data. You should also understand how to use the TFF platform for federated learning. In addition, you should understand the concept of TinyML and its requirements. Finally, you should have learned how TensorFlow Lite satisfies all the requirements of TinyML.

In the next chapter, we will learn about elastic model training and serving.