In BNM, loss value is negative

[guizilaile23]

Hi, first of all , thanks for the code and your paper, it's really excellent work.
And during the debug, i found the loss value is negative, is that right?
i debug the BNM in DA, the dataset was office31, source is amazon and target is dslr.
the "Transfer loss" will be -0.8 and classifier loss will be 0.02 after 2000 iterations.

and also, i found that if we simple calculate the classifier loss, the target acc will reach 100% near 1800 iteration, which means if we cut the BNM loss, it will be no harm to transfer the net from source to target.

am i missing something?
by the way, we are using pytorch1.9.0

We look forward to your reply. thanks again.

[cuishuhao]

1. negative loss value is right (I want to maximize the loss, the same with minimizing the opposite value of the loss).
2. Do you get 100% acc with"source is amazon and target is dslr"? If so, i think something must be wrong. If "D to W" or "W to D" is 100% acc, maybe it is caused by pytorch1.9.0.

[guizilaile23]

Sorry, i check the code, it's my mistake, have some bug about the datasets.
By the way, i fully understand your idea in paper, maximize the nuclear-norm, and the results is nearly the same with papers, it's just i don't know

but why not use 1/exp(-torch.mean(s_tgt))? i think it's have the same effect to maximize the nuclear-norm, and also make loss value positive.

i test 1/exp(-torch.mean(s_tgt)) as transfer loss, it's seems a little higher that directly use -torch.mean(s_tgt)

ie. office-home, from Real_World to Art, the target domain acc is 0.71982, and use -torch.mean(s_tgt), the target acc is 0.69551

[cuishuhao]

Good Try! I think you are better.