def _diag_fisher(self, cls_id, iteration, side_fc=False):
print("Training MAS model: Classifier %d" % (cls_id))
precision_matrices = {}
mse_criterion = nn.MSELoss()
mse_criterion = mse_criterion.to(device)
for n, p in copy.deepcopy(self.params).items():
p.data.zero_()
precision_matrices[n] = variable(p.data)
self.model.eval()
for batch_idx, (inputs, targets) in enumerate(self.dataset):
inputs, targets = inputs.to(device), targets.to(device)
#num_old_classes = args.nb_cl * iteration
#targets = targets - num_old_classes
if side_fc is True:
start_index = args.side_classifier * args.nb_cl * iteration
else:
start_index = args.nb_cl * iteration
self.model.zero_grad()
outputs = self.model(inputs, side_fc=side_fc)
i = cls_id - 1
Target_zeros = torch.zeros_like(
outputs[:, (start_index +
args.nb_cl * i):(start_index + args.nb_cl *
(i + 1))]).to(device)
loss_cls = mse_criterion(
outputs[:,
(start_index + args.nb_cl * i):(start_index +
args.nb_cl * (i + 1))],
Target_zeros)
loss_cls.backward()
for n, p in self.model.named_parameters():
if 'fc' not in n:
precision_matrices[n].data += torch.abs(p.grad.data) / len(
self.dataset)
if (batch_idx + 1) % 200 == 0:
print(batch_idx + 1)
precision_matrices = {n: p for n, p in precision_matrices.items()}
save_name_side = os.path.join(
ckp_prefix +
'WI/Weigtht_Importance_step_{}_K_{}_classifier_{}.pkl').format(
iteration, args.side_classifier, cls_id)
utils_pytorch.savepickle(precision_matrices, save_name_side)
return precision_matrices
svhn_data_copy = svhn_data.data
svhn_labels_copy = svhn_data.labels
# Launch the different runs
for n_run in range(args.nb_runs):
# Select the order for the class learning
order_name = "./checkpoint/{}_order_run_{}.pkl".format(args.dataset, n_run)
print("Order name:{}".format(order_name))
if os.path.exists(order_name):
print("Loading orders")
order = utils_pytorch.unpickle(order_name)
else:
print("Generating orders")
order = np.arange(args.num_classes)
np.random.shuffle(order)
utils_pytorch.savepickle(order, order_name)
order_list = list(order)
print(order_list)
start_iter = 0
for iteration in range(start_iter, int(args.num_classes / args.nb_cl)):
# Prepare the training data for the current batch of classes
actual_cl = order[range(iteration * args.nb_cl,
(iteration + 1) * args.nb_cl)]
indices_train_subset = np.array([
i in order[range(iteration * args.nb_cl,
(iteration + 1) * args.nb_cl)]
for i in Y_train_total
])
indices_test_subset = np.array([
i in order[range(0, (iteration + 1) * args.nb_cl)]
svhn_data_copy = svhn_data.data
svhn_labels_copy = svhn_data.labels
# Launch the different runs
for n_run in range(args.nb_runs):
# Select the order for the class learning
order_name = "./checkpoint/{}_order_run_{}.pkl".format(args.dataset, n_run)
print("Order name:{}".format(order_name))
if os.path.exists(order_name):
print("Loading orders")
order = utils_pytorch.unpickle(order_name)
else:
print("Generating orders")
order = np.arange(args.num_classes)
np.random.shuffle(order)
utils_pytorch.savepickle(order, order_name)
order_list = list(order)
print(order_list)
start_iter = 0
for iteration in range(start_iter, int(args.num_classes / args.nb_cl)):
# Prepare the training data for the current batch of classes(total class(100)/group class(20))
actual_cl = order[range(iteration * args.nb_cl,
(iteration + 1) * args.nb_cl)]
indices_train_subset = np.array([
i in order[range(iteration * args.nb_cl,
(iteration + 1) * args.nb_cl)]
for i in Y_train_total
])
indices_test_subset = np.array([
i in order[range(0, (iteration + 1) * args.nb_cl)]