Graph-Convolutional-Network
Why Convolution ??
Conventional Convolutional Neural Network
- Sharing kernel filter parameters
- Capture local area
Graph Convolutional Network
- Sharing parameters
- We use the same weight when computing the values of neighboring vertices in each vertex.
- Capture local area
- We only see neighboring vertices at each vertex.
Method
Implementation
Annotation
- A: Adjacent Matrix
- W: Weight Parameters