Link Prediction
Intermediate Google DeepMind, Meta+2
The problem, from first principles
Link prediction predicts missing or future edges in a graph. Given node embeddings h_u and h_v, score the likelihood of an edge. Common scorers: dot product, Euclidean distance, or MLP.
s(u,v)=huThvFrom theory to code
Implement:
link_prediction_score(u_emb, v_emb, method='dot')
Scores edge likelihood between two nodes.
Constraints
- u_emb, v_emb shape: (embedding_dim,).
- method in ('dot', 'euclidean', 'cosine').
- Return: scalar score.
Keywords
deep-learninggraph-neural-networkslink-prediction
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.