To overcome these shortcomings, a dynamic spatial–temporal subgraph convolutional network (DSTSGCN) is proposed for noncontact fault diagnosis. First, an edge-level dynamic graph convolutional network ...

The key challenge to build a pre-trained GNN on molecules is how to (1) generate well-distinguished graph-level representations and (2) automatically discover the functional groups without prior ...

Note: preprocessing is turned off in all experiment in our main paper. All shaDow-GNN are trained in the minibatch fashion. For each training batch, we first perform subgraph extraction, and then ...