Implicit vs Unfolded Graph Neural Networks (2111.06592v2)

Abstract: It has been observed that graph neural networks (GNN) sometimes struggle to maintain a healthy balance between the efficient modeling long-range dependencies across nodes while avoiding unintended consequences such oversmoothed node representations or sensitivity to spurious edges. To address this issue (among other things), two separate strategies have recently been proposed, namely implicit and unfolded GNNs. The former treats node representations as the fixed points of a deep equilibrium model that can efficiently facilitate arbitrary implicit propagation across the graph with a fixed memory footprint. In contrast, the latter involves treating graph propagation as unfolded descent iterations as applied to some graph-regularized energy function. While motivated differently, in this paper we carefully quantify explicit situations where the solutions they produce are equivalent and others where their properties sharply diverge. This includes the analysis of convergence, representational capacity, and interpretability. In support of this analysis, we also provide empirical head-to-head comparisons across multiple synthetic and public real-world node classification benchmarks. These results indicate that while IGNN is substantially more memory-efficient, UGNN models support unique, integrated graph attention mechanisms and propagation rules that can achieve SOTA node classification accuracy across disparate regimes such as adversarially-perturbed graphs, graphs with heterophily, and graphs involving long-range dependencies.