📗 -> 05/28/25: ECS189G-L24
🎤 Vocab
❗ Unit and Larger Context
GNN Outline
- Graph Neural Net vs other Deep Models
- GNN and Graph Embedding
- Graph Convolutional Network (GCN)
- GNN Training
- Graph Attention Network (GAT)
- GNN Applications
Quick Summary
Graph neural network introduction
- Graph vs image and sequential data
- GNN learning objectives
SGC operator and GCN - What is the SGC operator and how it works
- How to use SGC operator to build the GCN model
GAT - Potential problem with GCN in information aggregation
What is the GAT architecture GNN Applications - Some application examples
- More to be introduced in next class
✒️ -> Scratch Notes
Slide 13 - Graph Convolutional Net with “K” SGC Operator Layers
Simple Normalization Methods:
If you have an a matrix
Row Normalization -
Column Normalization -
Symmetric Normalized Adjacency Matrix:
Semi-supervised learning - A combination of supervised and unsupervised learning, some data has labels others do not
lot of skipping
GNN with Attention
- Larger
means higher attention - Attention function: e.g. inner product
GNN Implications:
- Graph Classification
- Brain Graphs (connectivity graphs), or classify as healthy or sick
- CV Scene Graphs, create object segmentations then use a GNN to create sentence outputs describing image
- NLP knowledge graph
Model Accuracy:
Similar as paper with the same dataset:
Citeseer - 67.9
Cora - 80.1
Pubmed - 78.9
NELL - 58.4
🧪 -> Refresh the Info
Did you generally find the overall content understandable or compelling or relevant or not, and why, or which aspects of the reading were most novel or challenging for you and which aspects were most familiar or straightforward?)
Did a specific aspect of the reading raise questions for you or relate to other ideas and findings you’ve encountered, or are there other related issues you wish had been covered?)
🔗 -> Links
Resources
- Put useful links here
Connections
- Link all related words