Supplementary Material for " Contrastive Graph Poisson Networks: Semi-Supervised Learning with Extremely Limited Labels " Sheng Wan

All experiments were conducted on a Linux server with a Tesla P40 GPU. The number of GA T layers was set as two. This work was done when Sheng was a research intern at JD Explore Academy. C. Gong is the corresponding author. We report the classification results of our CGPN framework and the baseline methods on the four datasets when five labels per class are available for training.