
Building Very Deep
Graph Neural Networks
for Representation
Learning on Graphs
Guohao Li
CS PhD Student @ KAUST

Building Very Deep Graph Neural Networks for Representation Learning on Graphs
Building Very Deep Graph Neural Networks for
Representation Learning on Graphs
4
Discussion:
To deep or not to deep
2
Making GCNs Go as
Deep as CNNs:
Message Aggregation
Functions;
Memory Efficiency
3
Designing GCNs
automatically:
Sequential Greedy
Architecture Search;
Latency Constrained;
1
Skip Connections and
Dilated Convolutions on
Graphs
Making GCNs Go as
Deep as CNNs:

Building Very Deep Graph Neural Networks for Representation Learning on Graphs
General Graphs:
● Social Networks
● Citation Networks
Lots of real-world applications need to deal with Non-Grid data
DeepGCNs.org
Graph data

Building Very Deep Graph Neural Networks for Representation Learning on Graphs
General Graphs:
● Social Networks
● Citation Networks
● Molecules
Lots of real-world applications need to deal with Non-Grid data
DeepGCNs.org
Graph data

Building Very Deep Graph Neural Networks for Representation Learning on Graphs
General Graphs:
● Social Networks
● Citation Networks
● Molecules
● Point Clouds
● 3D Meshes
● ...
Lots of real-world applications need to deal with Non-Grid data
DeepGCNs.org
Graph data