A Deep Graph Neural Networks Architecture Design: From Global Pyramid-like Shrinkage Skeleton to Local Topology Link Rewiring
This work addresses the problem of improving GNN expressivity and performance for researchers and practitioners working with graph data, offering an incremental advancement in architecture design.
This paper proposes a three-pipeline training framework for deep graph neural networks (GNNs) to improve expressivity, focusing on global model contraction, weight evolution, and link rewiring. The framework utilizes a pyramidal-like skeleton to address saddle points and rewires erroneous links based on network modularity, leading to significantly improved performance in node classification with faster convergence and enhanced robustness.
Expressivity plays a fundamental role in evaluating deep neural networks, and it is closely related to understanding the limit of performance improvement. In this paper, we propose a three-pipeline training framework based on critical expressivity, including global model contraction, weight evolution, and link's weight rewiring. Specifically, we propose a pyramidal-like skeleton to overcome the saddle points that affect information transfer. Then we analyze the reason for the modularity (clustering) phenomenon in network topology and use it to rewire potential erroneous weighted links. We conduct numerical experiments on node classification and the results confirm that the proposed training framework leads to a significantly improved performance in terms of fast convergence and robustness to potential erroneous weighted links. The architecture design on GNNs, in turn, verifies the expressivity of GNNs from dynamics and topological space aspects and provides useful guidelines in designing more efficient neural networks.