Module - Deep Learning on graphs (3)

Table of Contents

Graph embedding

0:00 Introduction
1:30 Graph embedding
2:43 How to represent graphs?
3:58 Why graph symmetries matter?
8:25 Invariant and equivariant functions
12:30 Message passing GNN
16:02 The many flavors of MGNN
20:00 Separating power
22:51 2-Weisfeiler-Lehman test
26:59 How powerful are MGNN
28:27 Empirical results
29:10 Graphs as higher order tensors
31:45 Invariant and equivariant linear operator
35:47 Invariant linear GNN
38:18 Folklore GNN

Slides

• slides

Post

• Invariant and equivariant layers with applications to GNN, PointNet and Transformers