So i wonder with all the multitude of parameters we’re dealing with while constructing neural nets (number of layers, dropout rates, learning rate, choice of functions to optimize and calculate loss, etc…) - could these tasks also be delegated to neural net - to come up with the most efficient and optimal neural net architecture for the given task?
I’m sure not the first one to ask this question.
Perhaps we can explore answers during Part 2 in February?
Gleb