Dual-Primal Graph Convolutional Networks

TL;DR

This paper introduces Dual-Primal Graph CNN, a novel architecture that performs convolution operations on both graphs and their duals, enabling learning of vertex and edge features with state-of-the-art results across various graph-based tasks.

Contribution

The paper presents a new dual-primal graph convolutional network architecture that extends existing models by incorporating dual graph operations for improved feature learning.

Findings

Achieved state-of-the-art results on citation network benchmarks.

Demonstrated effectiveness on graph-guided recommender systems.

Validated the model's generalization across multiple graph datasets.

Abstract

In recent years, there has been a surge of interest in developing deep learning methods for non-Euclidean structured data such as graphs. In this paper, we propose Dual-Primal Graph CNN, a graph convolutional architecture that alternates convolution-like operations on the graph and its dual. Our approach allows to learn both vertex- and edge features and generalizes the previous graph attention (GAT) model. We provide extensive experimental validation showing state-of-the-art results on a variety of tasks tested on established graph benchmarks, including CORA and Citeseer citation networks as well as MovieLens, Flixter, Douban and Yahoo Music graph-guided recommender systems.