Data-Driven Learning of Geometric Scattering Networks
This work addresses limitations in GNNs for graph classification and exploration tasks, offering an incremental improvement through adaptive wavelet tuning.
The authors tackled the problem of learning longer-range graph relations in graph neural networks (GNNs) by proposing a learnable geometric scattering (LEGS) module, which resulted in simplified architectures with significantly fewer parameters and demonstrated predictive performance on graph classification benchmarks.
We propose a new graph neural network (GNN) module, based on relaxations of recently proposed geometric scattering transforms, which consist of a cascade of graph wavelet filters. Our learnable geometric scattering (LEGS) module enables adaptive tuning of the wavelets to encourage band-pass features to emerge in learned representations. The incorporation of our LEGS-module in GNNs enables the learning of longer-range graph relations compared to many popular GNNs, which often rely on encoding graph structure via smoothness or similarity between neighbors. Further, its wavelet priors result in simplified architectures with significantly fewer learned parameters compared to competing GNNs. We demonstrate the predictive performance of LEGS-based networks on graph classification benchmarks, as well as the descriptive quality of their learned features in biochemical graph data exploration tasks.