We saw how a neural Turing machine stores and retrieves information from the memory and how it uses different addressing mechanisms, such as location-based and content-based addressing, for reading and writing information. We also learned how to implement NTM using TensorFlow to perform copy tasks. Then, we learned about MANN and how MANN differs from NTM. We also learned how MANN uses the least recently used access method to overcome the shortcomings of NTM.

In the next chapter, we will learn about Model Agnostic Meta Learning (MAML) and how it is used in a supervised and reinforcement learning setting.