In this part, we will take part in a real R&D process to build a robust and relevant Recurrent Neural Network. Here is the plan of attack:
We will study the theory and get the Intuition of RNNs.
We will start by building a simple RNN, our first attempt.
We will observe the results to identify possible issues and ways of improvement, so that eventually this simple RNN will be well improved in the last section.
You will have a first Homework to already practice with that simple RNN.
We will learn how to evaluate a RNN model, and more generally a Regression model.
After having clearly identified the issues and improvement solutions, we will make several more robust and relevant RNNs. This will be the final big Homework Challenge of this course. The solution will be provided in a zip folder with all the final models and their plotted results.