Skip to main content

Table 1 Comparison of Complexity of GraphSAGE and Graph Convolutional Networks (GCN) (Adapted from [67]) n is the total number of nodes. m is the total number of edges. K is the number of layers. s is the batch size. r is the number of neighbours being sampled for each node

From: Tackling racial bias in automated online hate detection: Towards fair and accurate detection of hateful users with geometric deep learning

Complexity

GCN

GraphSAGE

Time

\(O(Kmd + Knd^{2})\)

\(O(r^{K}nd^{2})\)

Memory

\(O(Knd + Kd^{2})\)

\(O(sr^{K}d + Kd^{2})\)