Question 15
An end-to-end approach doesn’t require that we hand-design useful features, it only requires a large enough model. True/False?
I find this question poorly written since in the lecture we were told repeatedly that end to end learning requires a lot of data to have a chance of working well. Yet the answers don’t take this into account.
Hi @abelian_group_chen ,
it is nice to have you here and thank you for the question!
Though I understand your concern, that could be extended to many more parts of “what is required”, right? I believe the question refers to the feature extraction process and therefore I find the formulation quite accurate.
Thanks for the reply. If the wording of the question was changed to “one of the requirements is to have a large enough model” then I would agree. In it’s current wording I find it quite misleading given the T/F nature of the question. Perhaps this question would be better framed as a multiple choice question?