Abstract
The scattering transform is a multilayered wavelet-based deep learningarchitecture that acts as a model of convolutional neural networks. Recently,several works have introduced generalizations of the scattering transform fornon-Euclidean settings such as graphs. Our work builds upon these constructionsby introducing windowed and non-windowed graph scattering transforms based upona very general class of asymmetric wavelets. We show that these asymmetricgraph scattering transforms have many of the same theoretical guarantees astheir symmetric counterparts. This work helps bridge the gap between scatteringand other graph neural networks by introducing a large family of networks withprovable stability and invariance guarantees. This lays the groundwork forfuture deep learning architectures for graph-structured data that have learnedfilters and also provably have desirable theoretical properties.