Closed hooseok closed 3 years ago
Hi, thanks for your interest. "CE(Balanced)" means CE with class-balanced sampling, which corresponds to "Resample" as for args.train_rule
. You can also choose "Reweight", which means re-weighting the loss for each class according to # of samples, by changing args.train_rule
.
https://github.com/YyzHarry/imbalanced-semi-self/blob/16d8f02264d9e16602d1a47acc43053b6bb007c4/train.py#L31-L32
I see the Self-supervised pretrained learning (SSP). There are many models in SSP.
Where can I setting the CB in train.py code? In my opinion, per_cls_weights seems to set a uniform or balance. Does the CB setting mean 'Reweight' in args.train_rule?