def load_test_data(self, m_5=6000, k=1000):
test_bgs_data = model_utils.load_test_bg_data(self.path_dataset)
test_signal_data = model_utils.load_test_signal_data(self.path_dataset,
m_5=m_5,
k=k)
factor = -1 * np.log(0.01)
#
# factor = 1
norm_test_bgs_data = model_utils.normalize(test_bgs_data, factor)
norm_test_signal_data = model_utils.normalize(test_signal_data, factor)
reshape_norm_test_bg = np.reshape(norm_test_bgs_data, (
norm_test_bgs_data.shape[0],
norm_test_bgs_data.shape[1],
norm_test_bgs_data.shape[2],
1))
reshape_norm_train_bg = np.reshape(norm_test_signal_data, (
norm_test_signal_data.shape[0],
norm_test_signal_data.shape[1],
norm_test_signal_data.shape[2],
1))
return reshape_norm_test_bg, reshape_norm_train_bg
def load_train_data(self):
train_data = model_utils.load_train_data(self.path_dataset)
test_bgs_data = model_utils.load_test_bgs_data(self.path_dataset)
factor = -1 * np.log(0.01)
norm_train_data = model_utils.normalize(train_data, factor)
norm_test_bgs_data = model_utils.normalize(test_bgs_data, factor)
return norm_train_data, norm_test_bgs_data
def load_test_data(self, signal_id=1):
test_bgs_data = model_utils.load_test_bgs_data(self.path_dataset)
test_signal_data = model_utils.load_test_signal_data(
self.path_dataset, signal_id=signal_id)
factor = -1 * np.log(0.01)
norm_test_bgs_data = model_utils.normalize(test_bgs_data, factor)
norm_test_signal_data = model_utils.normalize(test_signal_data, factor)
return norm_test_bgs_data, norm_test_signal_data
def load_train_data(self):
train_data = model_utils.load_train_data(self.path_dataset)
test_bgs_data = model_utils.load_test_bgs_data(self.path_dataset)
train_shape = (self.dataset_config['TRAIN_SIZE'], self.shape)
test_bgs_shape = (self.dataset_config['TEST_BACKGROUND_SIZE'], self.shape)
factor = -1 * np.log(0.01)
norm_train_data = model_utils.normalize(train_data.reshape(train_shape), factor)
norm_test_bgs_data = model_utils.normalize(test_bgs_data.reshape(test_bgs_shape), factor)
return norm_train_data, norm_test_bgs_data
def load_test_data(self, signal_id=1):
test_bgs_data = model_utils.load_test_bgs_data(self.path_dataset)
test_signal_data = model_utils.load_test_signal_data(self.path_dataset, signal_id=signal_id)
test_bgs_shape = (len(test_bgs_data), self.shape)
test_signal_shape = (len(test_signal_data), self.shape)
print(test_signal_data.shape)
factor = -1 * np.log(0.01)
norm_test_bgs_data = model_utils.normalize(test_bgs_data.reshape(test_bgs_shape), factor)
norm_test_signal_data = model_utils.normalize(test_signal_data.reshape(test_signal_shape), factor)
return norm_test_bgs_data, norm_test_signal_data
def forward(self, input):
gpu_ids = None
if isinstance(input.data, torch.cuda.FloatTensor) and self.ngpu > 0:
gpu_ids = range(self.ngpu)
output = nn.parallel.data_parallel(self.main, input, gpu_ids)
else:
output = self.main(input)
output = output.view(output.size(0), -1)
if self.noise == 'sphere':
output = utils.normalize(output)
return output
def load_train_data(self):
train_bg = model_utils.load_train_bg_data(self.path_dataset)
test_bg = model_utils.load_test_bg_data(self.path_dataset)
factor = -1 * np.log(0.01)
norm_train_bg = model_utils.normalize(train_bg, factor)
norm_test_bg = model_utils.normalize(test_bg, factor)
# norm_train_bg = train_bg
# norm_test_bg = test_bg
reshape_norm_train_bg = np.reshape(norm_train_bg, (
norm_train_bg.shape[0],
norm_train_bg.shape[1],
norm_train_bg.shape[2],
1))
reshape_norm_test_bg = np.reshape(norm_test_bg, (
norm_test_bg.shape[0],
norm_test_bg.shape[1],
norm_test_bg.shape[2],
1))
return reshape_norm_train_bg, reshape_norm_test_bg
