domain label are different from paper

[jhzhang19]

Thank you for your great work. I find a question in loss function: the domain_s label is 1 and domain_t label is 0 in paper, but in the code domain label are 0 and 1. In addition,the loss function in paper is contradictory

[Shuntw6096]

I think that the domain_s label is 1 and domain_t label is 0 the paper mentioned is for backbone (generator), because the two alignment networks(discriminator) connect to backbone through gradient reverse, for backbone, optimization direction is minimum, source domain label is 1 we want for backbone, but discriminator use reverse optimization direction to train, so use flipped label to train), same as GAN training.(if I understand adversarial learning correctly)

[a21401624]

Thank you for your great work. I find a question in loss function: the domain_s label is 1 and domain_t label is 0 in paper, but in the code domain label are 0 and 1. In addition,the loss function in paper is contradictory

I read the paper yesterday and I think I have found the same issue. The formulation (8) and (9) in the paper are inverse, if the domain_s label is 1 and domain_t label is 0. I think the question is evident.