def get_models(input_channels=train_ds.input_channels, seq_len=train_ds.seq_len,
pred_out_size=pred_target_values.shape[0], disc_out_size=disc_target_values.shape[0],
phys_cols=phys_cols, act_cols=act_cols):
predictor = M.PredictorConv(input_channels=input_channels, seq_len=seq_len, output_size=pred_out_size,
physNodes=phys_cols)
#M.load_classifier_state2(predictor, "Predictor")
predictor.to(DEVICE)
pred_optim = M.get_optimizer(predictor,)
discriminator = M.DiscriminatorConv(input_channels=input_channels, seq_len=seq_len, output_size=disc_out_size,
physNodes=phys_cols+act_cols)
#M.load_classifier_state2(discriminator,"Discriminator")
discriminator.to(DEVICE)
disc_optim = M.get_optimizer(discriminator,)
return predictor, pred_optim, discriminator, disc_optim
accAct += np.abs((y == yp).data.numpy()).sum()
except TypeError:
accAct += np.abs((y.cpu() == yp.cpu()).data.numpy()).sum()
accAct = accAct / test_ds.length
print("Accuracy: {}".format(accAct))
PredicActAccu.append(accAct)
# Defining model Predicting activities.
activities = np.unique(train_ds.sensitive)
phys_shape = train_ds.phy_data.shape[1]
#Parameters
model = M.DiscriminatorConv(input_channels=train_ds.input_channels,
seq_len=train_ds.seq_len,
output_size=activities.shape[0],
physNodes=phys_shape)
# Send model on GPU or CPU
model.to(device)
# Loss
# loss = torch.nn.NLLLoss()
# loss = Cl.NLLLoss()
loss = Cl.BalancedErrorRateLoss(targetBer=0, device=device)
# loss = Cl.AccuracyLoss(device=device)
# Training procedure
max_epochs = 200
losses = []
#t_key = "act"
t_key = "sens"
print("Starting Training")
for i in tqdm.tqdm(range(max_epochs)):