Course 4, week 2, assignment 2, exercise 1: my suggestion

Hello dear DeepLearning.AI support team.

I did exercise 1 of the second home work of week 2, course 4. But I had real problems understanding it. I would like it to be redesigned.

Here are the problems I experienced and suggestions to mitigate them.

  1. Problem: I didn’t understand the statement of the exercise (actually - even having passed the tests - I still don’t understand what I did in the exercise).

       1.1 Suggestion. Explain what the model is that we are talking about. Explain why data augmentation is a part of the main model and not a part of preprocessing (i.e. I think that preprocessing should result in an input vector X that is fed into the main model, but you create some additional preprocessing model). Explain why preprocessing is done in layers. Draw those layers and how they are integrated into the main model.
  2. Problem: in week 1 we did use Sequential but we didn’t use .add method. We created all the layers in one line of code without using .add. So, in this exercise I created the layers in one line of code like we did in week 1 and then had a really hard time understanding what I was supposed to do with those 2 lines with the .add method.

       2.1. Suggestion. Briefly explain the ways layers can be created in Sequential.

Also, just so you know. I cannot understand what is written in the paragraph before exercise 1. Dr. Ng mentioned something related to that, but it is not enough to understand that paragraph. Can you rewrite it please (with pictures preferably).

Thank you in advance.


Thanks for your suggestions.