Motif-based graph representation learning with application to chemical molecules
We designed a novel convolution module for graph representational learning on molecules with an efficient pretraining strategy, enabling the capture of local structural and semantic information from graph motifs.
Jan 11, 2023