Fused Gromov-Wasserstein Graph Mixup for Graph-level Classifications Xinyu Ma

GNNs rely on parameterized nonlinear transformations to process graph signals, which can be efficiently encoded and serve as crucial inputs for downstream predictions.