labels_dict,
window_sz,
pca=False,
noise_percentage=0.2,
noise=False)
datasets = LoadData.load_data_multi_samples(dataset_matrix_r,
label_vector_r, n_components,
n_classes, samples)
train_set_x, train_set_y = datasets[0][0], datasets[0][1]
test_set_x, test_set_y = datasets[1][0], datasets[1][1]
train_set_x, scaler = PCA.scale_fit(train_set_x)
test_set_x = PCA.scale_transform(test_set_x, scaler)
batches = LoadData.batch_creation(train_set_x, train_set_y, batch_size=65)
del datasets, dataset_matrix, label_vector, dataset_matrix_r, label_vector_r, dataset_mat, dataset_dict, labels_mat, labels_dict
x_0 = set_tensor_matricization(train_set_x, mode=0)
x_1 = set_tensor_matricization(train_set_x, mode=1)
x_2 = set_tensor_matricization(train_set_x, mode=2)
W_0 = torch.empty(hidden, rank, window_sz) # Initialize with xavier
W_1 = torch.empty(hidden, rank, window_sz)
W_2 = torch.empty(hidden, rank, n_components)
V = torch.empty(n_classes, hidden)
b = torch.zeros(n_classes)
nn.init.xavier_uniform_(W_0)
nn.init.xavier_uniform_(W_1)
