Week 11 — Current Research Directions

Where the field is heading: scale, expressivity, foundations.

Lecture

Scaling laws for GNNs · the over-smoothing and over-squashing problems · positional encodings on graphs · graph transformers · the geometric foundation models question.

Read before the lecture

• Rampášek et al., *Recipe for a General, Powerful, Scalable Graph Transformer* (NeurIPS 2022)

Recitation — paper discussion

Tönshoff et al., *Where Did the Gap Go? Reassessing the Long-Range Graph Benchmark* (2023) (paper)

Come ready to argue one side of each:

• Are claimed long-range gains real or evaluation artifacts?

Reference text for this week: chapter 11 of the bilingual notes — EN PDF · FR PDF.