bilevel_augment
bilevel_augment copied to clipboard
This method is an online method?
First of all, thank you very much for the excellent work. My question is, does this method involve retraining? That is, fix the parameters of the augmentation network finally obtained, and then retrain the classifier with the searched augmentation network using all the training data.
In other words, once the classifier or data set is changed, does this method need to be run again?