I got a question about end to end learning:
An end-to-end approach doesn’t require that we hand-design useful features, it only requires a large enough model. True/False?
The official answer is True. But I think the answer is false because of the 2nd part of the sentence as it says: it only requires large enough model. However, the hand designed features are not needed but not only a large model requires but also large amount of sample data.
This question has come up before. I think you’re just over-interpreting the question. It’s never the case that just the architecture of your model is enough to guarantee success. You always need enough training data, you need the compute resources to run the training, you need to make sure the power grid stays up long enough to complete your training and on and on … But it’s not asking about any of those other requirements: it’s specifically asking about whether hand engineered features are required or not versus a complex enough model.