This is a repository for the JMLR paper Limiting Over-Smoothing and Over-Squashing of Graph Message Passing by Deep Scattering Transforms

We propose the DSMP model in this paper, which is a discriminatively trained, multi-layer graph neural network model designed to overcome over-smoothing, over-squashing and instabilty issues.

By harnessing spectral transformation, the DSMP model aggregates neighboring nodes with global information, thereby enhancing the precision and accuracy of graph signal processing. We provide theoretical proofs demonstrating the DSMP’s effectiveness in mitigating these issues under specific conditions. Additionally, we support our claims with empirical evidence and thorough frequency analysis, showcasing the DSMP’s superior ability to address instability, over-smoothing, and oversquashing.

python3 smp_noabs_molhiv arg_train # if test for the first time, use arg_train action, else use ray_train for multiple tests.

python3 smp_graphlevel_noabs arg_train --dataname COLLAB # need to specify the name of the dataset

python3 smp_nodeclass_noabs arg_train --dataname wisconsin # need to specify the name of the dataset