Skip to main content

Advertisement

Table 2 Summary of experimental results

From: Learning compact graph representations via an encoder-decoder network

Method Dataset   
  HIV NCI81 NCI83
ECFP 68.30% 68.90% 62.06%
NeuralFPS 67.48% 65.24% 59.91%
DeepWalk 69.90% 68.00% 63.89%
3-GK 61.78% 53.34% 53.01%
SP 59.10% 53.02% 56.26%
Skip-graph (ours) 72.77% 69.98% 63.80%
  1. The best result(s) per dataset are highlighted.