Adjacency based embeddings

File structure:

1. MAS.py: implementation of the MAS algorithm
2. TAS.py: implementation of TAS algorithm
3. main.ipynb: Sample notebook demonstrating execution of TAS and MAS to get the embeddings

How to run?

Please refer to main.ipynb to check the execution process. We have provided the code to execute MAS (implemented in MAS.py) and TAS (implemented in TAS.py).

Data format

Please provide the data as a pickle file in the below format:

[g,
[train_X, train_Y],
[val_X, val_Y],
[test_X, test_Y]]

As a sanity check, plese ensure that the pickle file can be loaded using the code below:

with open(DATA_PATH, "rb") as f:
    (
        g,
        [train_X, train_y],
        [val_X, val_y],
        [test_X, test_y],
    ) = pickle.load(f)

Here,

• g is a networkx.classes.graph.Graph object containing the graph
• train_X, val_X and test_X are lists containing train, val and test splits respectively. Every element of the list represents the node ID of a node in the respective split
• train_Y, val_Y and test_Y are lists containing the node labels for the respective splits.