Brain Signal Classification via Learning Connectivity Structure

Brain Signal Classification via Learning Connectivity Structure

Soobeom Jang, etc.

Proposed Method

Problem statement and notations

$$X \in \mathbb{R}^{N \times T}$$

graph structure $$G = (V, {E}{k=1}^K, {W^k}{k=1}^K )$$。其中$$V={vi}{i=1}^N$$表示vertices，$$E^k \in {0, 1}^N$$表示第k层graph layer每个edge是否存在，而$$W^k \in \mathbb{R}^{N \times N}$$表示的是第k层graph layer上任意两个node之间的权重，且所有权重在0到1之间。

$$K$$是一个hyperparameter。

Proposed Model

Appendix

不太理解这句话

This method is developed only for transductive learning, while our model is for inductive learning.