Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs

TLDR

Graph-structured data enable a variety of prediction problems. The study generalizes convolution to arbitrary graphs without spectral methods, enabling handling of graphs with varying size and connectivity. The method conditions filter weights on edge labels and uses graph coarsening to build deep neural networks for graph classification. The approach achieves state‑of‑the‑art results on point‑cloud classification and outperforms existing deep learning methods on a graph‑classification dataset.

Abstract

A number of problems can be formulated as prediction on graph-structured data. In this work, we generalize the convolution operator from regular grids to arbitrary graphs while avoiding the spectral domain, which allows us to handle graphs of varying size and connectivity. To move beyond a simple diffusion, filter weights are conditioned on the specific edge labels in the neighborhood of a vertex. Together with the proper choice of graph coarsening, we explore constructing deep neural networks for graph classification. In particular, we demonstrate the generality of our formulation in point cloud classification, where we set the new state of the art, and on a graph classification dataset, where we outperform other deep learning approaches.