This module normalizes positive scalar edge weights on a graph following the form in
GCN <https://arxiv.org/abs/1609.02907>
按照GCN论文里面提出的 normalize 方法对正向边处理。
FomulaThe module returns the normalized weight e j i c j i \frac{e_{ji}} {c_{ji}} cjieji
根据不同的配置,给出上述两个变量的计算公式:
Mathematically, setting
norm='both'
yields the following normalization term:
具体的数学公式是:
-
norm=both
c j i = ( ∑ k ∈ N ( j ) e j k ∑ k ∈ N ( i ) e k i ) c_{ji} = (\sqrt{\sum_{k\in\mathcal{N}(j)}e_{jk}}\sqrt{\sum_{k\in\mathcal{N}(i)}e_{ki}}) cji=(k∈N(j)∑ejk k∈N(i)∑eki ) -
norm='right'
yields the following normalization term:
c j i = ∑ k ∈ N ( i ) e k i c_{ji} = \sum_{k\in\mathcal{N}(i)}e_{ki} cji=k∈N(i)∑eki
where e j i e_{ji} eji is the scalar weight on the edge from node
j
to nodei
.
e j i e_{ji} eji 是节点j
到节点i
的权重标量; N i N_i Ni代表的就是节点i的邻居