Graph Neural Networks

1 min
  1. Why must neighbor aggregation be order-invariant?
  2. What does a node see after two message-passing layers?
  3. How can a future edge create a leak?
  4. What is over-smoothing?
  5. [E] What control justifies a GNN against a tabular model?

Answers. 1. A graph has no canonical order of neighbors. 2. Up to two hops. 3. It transmits information unavailable at prediction time. 4. Node representations become too similar. 5. Compare, on the same split, a structure-free baseline.

Activity — Manual message passing

On a four-node graph, assign a scalar value to each, compute an average aggregation of neighbors then an update. Redo after permuting neighbor order. Success: the output stays identical and the split forbids any future edge.

Preview — the rest of the lesson is for enrolled readers.

Already enrolled with a code?

Your access is tied to your account, not to this link. Sign in with the same email you used in class: your course is waiting, no need to enter the code again.

Sign inNo account yet? Create one
This lesson is part of the “Modern Deep Dives” module

The first modules of the course are open to everyone. For the rest you have three options: buy this course once and for all, subscribe, or enter the code handed out in class.

Are you a student on this course?

The code is tied to your account: sign in or create an account and it will be applied automatically when you come back.

No account yet? Create one