GraphNorm: A Principled Approach to Accelerating
               Graph Neural Network Training

        Tianle Cai * 1 2 Shengjie Luo * 3 4 5 Keyulu Xu 6 Di He 7 Tie-Yan Liu 7 Liwei Wang 3 4

             Abstract                  2017). Empirically, GNNs have succeeded in a variety of
   Normalization is known to help the optimization        tasks such as computational chemistry (Stokes et al., 2020),
   of deep neural networks. Curiously, different         recommendation systems (Ying et al., 2 ...