Skip to main content

Table 2 For each dataset, the mean classification accuracy plus standard deviation of 10-fold cross validation for each graph classification model are displayed

From: An end-to-end graph convolutional kernel support vector machine

Model MUTAG PTC_MR BZR_MD PTC_FM COX2
Graphlet 0.86 ±0.05 0.54 ±0.08 0.61 ±0.12 0.57 ±0.07 0.75 ±0.06
Shortest path 0.82 ±0.06 0.55 ±0.08 0.68 ±0.07 0.55 ±0.07 0.77 ±0.06
Vertex histogram 0.85 ±0.05 0.58 ±0.07 0.70 ±0.08 0.58 ±0.06 0.75 ±0.06
Weisfeiler Lehman 0.71 ±0.06 0.59 ±0.07 0.60 ±0.07 0.63 ±0.08 0.65 ±0.08
Pyramid match 0.86 ±0.06 0.54 ±0.06 0.62 ±0.07 0.57 ±0.09 0.73 ±0.06
GCN 0.73 ±0.06 0.57 ±0.03 0.68 ±0.06 0.61 ±0.08 0.79 ±0.03
GCNWithJK 0.73 ±0.07 0.57 ±0.05 0.68 ±0.07 0.62 ±0.08 0.78 ±0.02
GIN 0.82 ±0.07 0.54 ±0.05 0.62 ±0.09 0.57 ±0.07 0.80 ±0.04
GIN0 0.85 ±0.04 0.57 ±0.08 0.63 ±0.13 0.59 ±0.05 0.77 ±0.04
GINWithJK 0.83 ±0.07 0.55 ±0.07 0.61 ±0.15 0.60 ±0.04 0.80 ±0.06
GIN0WithJK 0.83 ±0.06 0.54 ±0.07 0.63 ±0.10 0.59 ±0.06 0.82 ±0.05
GraphSAGE 0.72 ±0.06 0.57 ±0.08 0.68 ±0.09 0.61 ±0.06 0.78 ±0.01
GraphSAGEWithJK 0.71 ±0.09 0.56 ±0.04 0.68 ±0.09 0.60 ±0.07 0.77 ±0.01
DiffPool 0.84 ±0.12 0.57 ±0.03 0.69 ±0.07 0.61 ±0.06 0.77 ±0.02
GlobalAttentionNet 0.74 ±0.07 0.56 ±0.05 0.67 ±0.06 0.63 ±0.06 0.77 ±0.01
Set2SetNet 0.73 ±0.07 0.56 ±0.03 0.68 ±0.10 0.62 ±0.06 0.78 ±0.04
SortPool 0.75 ±0.11 0.59 ±0.08 0.66 ±0.09 0.60 ±0.08 0.77 ±0.01
Proposed model 0.87 ±0.06 0.60 ±0.08 0.62 ±0.10 0.62 ±0.06 0.79 ±0.05
  1. Boldface indicates best performing model