Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs

TL;DR

This paper introduces a novel convolutional approach for graph-structured data that dynamically conditions filters on edge labels, enabling effective deep learning for graph and point cloud classification.

Contribution

It generalizes convolution to arbitrary graphs with edge-conditioned filters, avoiding spectral methods, and demonstrates state-of-the-art results in point cloud and graph classification.

Findings

Achieved state-of-the-art accuracy in point cloud classification.

Outperformed existing methods on a graph classification dataset.

Demonstrated the flexibility of edge-conditioned filters in various graph tasks.

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. The source code is available at https://github.com/mys007/ecc