Semi-Supervised Classification with Graph Convolutional Networks

$H^{(l+1)}=\sigma ( \tilde{D} ^ {-\frac{1}{2}} \tilde{A} \tilde{D} ^ {-\frac{1}{2}} H^{(l)} W^{(l)})$

• ((62d9fcee-0982-4fa7-a6db-cb396fd0438b))
  • 避免顶点的度越大，学到的表示越大
  • ((62dc1e2b-dc16-42c3-99fc-fa14921059a8))
  • ((62dc1e26-5ca8-4302-9d27-ce283a33186b))
  • $\tilde{A}=A+I_N$
• $H^{(l+1)}=\sigma\left(\tilde{A} H^{(l)} W^{(l)}\right)$
  • $\tilde{A}$ 矩阵 nn，$H^{(l)}$ 矩阵 nm，$W$ 矩阵 mu，$H^{(l+1)}$ 矩阵 nu
  • $\tilde{A} H^{(l)}$ 考虑节点本身和邻居的信息
  • {:height 223, :width 448}

网络回响