KDD'23 Tutorial: 大规模 GNN 的过去和未来

https://sites.google.com/ncsu.edu/gnnkdd2023tutorial/home?authuser=0

Presenter

Outline

1. Introduction of GNNs (20minutes)

(a) Foundations and Applications of GNNs

(b) Scalability Challenges of Large-Scale GNNs

2. Classic Approaches for Scaling GNNs (50minutes)

(a) Sampling Methods

(b) Decoupling Methods

(c) Distributed Methods

Break: 10 minutes

3. Emerging Techniques for Scaling GNNs (50minutes)

(a) Lazy Graph Propagation

(b) Alternating Training

(c) Layer-wiseTraining

(d) Graph Condensation

(e) Graph Distillations

(f) GNN Pre-training

(g) GNN Pruning

4. Evaluation and Comparison of Scalable GNNs (20minutes)

5. Large-scale Real-world Applications (20 minutes)

6. Summary and Future Directions (10minutes)