I have an adjacency matrix (256,256), features (34800,256) and labels (34800,) which has [0,1,2,3]. I want to implement a model using DGL to perform classification with 4 classes output but I cant figure out a starting point.
Any pointers?
This definitely will guide me however I still do not understand the masking concept. I have tried to breakdown the one used in the tutorial where the authors have a method and use the labels.shape[0] value, but I still cannot understand the concept. This is the authors’ method and mask variables. The shape of the dataset is (2708,1433) and the labels.shape[0] is (2708,). What is the concept behind the mask? Where do the values 140, 200, 500 and 1500 come from?
def _sample_mask(idx, l):
"""Create mask."""
mask = np.zeros(l)
mask[idx] = 1
return mask
The masks
train_mask = _sample_mask(range(140), labels.shape[0])
val_mask = _sample_mask(range(200, 500), labels.shape[0])
test_mask = _sample_mask(range(500, 1500), labels.shape[0])
2 Likes