Tong Alexander, Wenkel Frederick, Macdonald Kincaid, Krishnaswamy Smita, Wolf Guy
Yale University, Dept. of Comp. Sci., New Haven, CT, USA.
Université de Montréal, Dept. of Math. & Stat.; Mila - Quebec AI Institute, Montreal, QC, Canada.
IEEE Int Workshop Mach Learn Signal Process. 2021 Oct;2021. doi: 10.1109/mlsp52302.2021.9596169. Epub 2021 Nov 15.
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.
我们提出了一种新的图神经网络（GNN）模块，该模块基于最近提出的几何散射变换的松弛形式，它由一系列图小波滤波器组成。我们的可学习几何散射（LEGS）模块能够对小波进行自适应调谐，以促使带通特征出现在学习到的表示中。与许多常用的GNN相比，将我们的LEGS模块整合到GNN中能够学习到更长距离的图关系，常用的GNN通常依靠通过邻居之间的平滑度或相似性来编码图结构。此外，与竞争的GNN相比，其小波先验导致架构简化，学习参数显著减少。我们展示了基于LEGS的网络在图分类基准上的预测性能，以及它们在生化图数据探索任务中学习到的特征的描述质量。
