In the Week 1 video where Robert talks about the search spaces. It kind of went over my head what we were trying to do in there. The nodes, macro and micro structures.
Can anyone help me with a simple/practical explanations or some resources to get some fair understanding around the same?
Thanks,
Yash
Hi Yash,
he describes the general idea there that one can automate the choice of hyperparameters, which you normally set before running the modelling process and that determine the model structure and process. As one does not normally know the ideal choice in advance, one needs to have a strategy to optimize them, as one cannot just simply try out the full range of combinations. And therefore one needs to have a strategy there to optimize those.
Some resources that might help:
https://en.wikipedia.org/wiki/Hyperparameter_(machine_learning)
Hope it helps,
Nils