Can anyone help with understanding how to best do #modelselection in the context of #neuralnetworks ? I'm trying to understand how to reduce #bias due to the selection of a particular test set.

More details here

https://stats.stackexchange.com/q/620547/582