def forward(self, source, target, s_label):
source = self.feature_layers(source)
if bottle_neck:
source = self.bottle(source)
s_pred = self.cls_fc(source)
if self.training ==True:
target = self.feature_layers(target)
if bottle_neck:
target = self.bottle(target)
t_label = self.cls_fc(target)
#原始的lmmd
loss = mmd.lmmd(source, target, s_label, torch.nn.functional.softmax(t_label, dim=1),num_classes = self.num_classes)
#混合核lmmd
#loss = mklmmd.mix_poly_rbf(source, target, s_label, torch.nn.functional.softmax(t_label, dim=1))
else:
loss = 0
return s_pred, loss
def forward(self, source, target, s_label):
source = self.feature_layers(source)
if bottle_neck:
source = self.bottle(source)
s_pred = self.cls_fc(source)
if self.training == True:
target = self.feature_layers(target)
if bottle_neck:
target = self.bottle(target)
t_label = self.cls_fc(target)
loss_c = mmd.lmmd(source, target, s_label,
torch.nn.functional.softmax(t_label, dim=1))
loss_m = self.mmd_marginal(source, target)
mu = proxy_a_distance.estimate_mu(
source.detach().cpu().numpy(),
s_label.detach().cpu().numpy(),
target.detach().cpu().numpy(),
torch.max(t_label, 1)[1].detach().cpu().numpy())
else:
loss_c, loss_m, mu = 0
return s_pred, loss_c, loss_m, mu