DiffWire: Inductive Graph Rewiring via the Lovász Bound

15 Jun 2022  ·  Adrian Arnaiz-Rodriguez, Ahmed Begga, Francisco Escolano, Nuria Oliver ·

Graph Neural Networks (GNNs) have been shown to achieve competitive results to tackle graph-related tasks, such as node and graph classification, link prediction and node and graph clustering in a variety of domains. Most GNNs use a message passing framework and hence are called MPNNs. Despite their promising results, MPNNs have been reported to suffer from over-smoothing, over-squashing and under-reaching. Graph rewiring and graph pooling have been proposed in the literature as solutions to address these limitations. However, most state-of-the-art graph rewiring methods fail to preserve the global topology of the graph, are neither differentiable nor inductive, and require the tuning of hyper-parameters. In this paper, we propose DiffWire, a novel framework for graph rewiring in MPNNs that is principled, fully differentiable and parameter-free by leveraging the Lov\'asz bound. The proposed approach provides a unified theory for graph rewiring by proposing two new, complementary layers in MPNNs: CT-Layer, a layer that learns the commute times and uses them as a relevance function for edge re-weighting; and GAP-Layer, a layer to optimize the spectral gap, depending on the nature of the network and the task at hand. We empirically validate the value of each of these layers separately with benchmark datasets for graph classification. We also perform preliminary studies on the use of CT-Layer for homophilic and heterophilic node classification tasks. DiffWire brings together the learnability of commute times to related definitions of curvature, opening the door to creating more expressive MPNNs.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Node Classification Actor CT-Layer Accuracy 31.98 # 43
Node Classification Actor CT-Layer (PE) Accuracy 29.35 # 48
Node Classification Citeseer CT-Layer Accuracy 66.71 # 65
Node Classification Citeseer CT-Layer (PE) Accuracy 72.26 # 45
Graph Classification COLLAB CT-Layer Accuracy 69.87% # 28
Graph Classification COLLAB DiffWire Accuracy 72.24% # 26
Graph Classification COLLAB GAP-Layer (Ncut) Accuracy 65.89% # 32
Graph Classification COLLAB GAP-Layer (Rcut) Accuracy 64.47% # 34
Node Classification Cora CT-Layer (PE) Accuracy 83.66% # 37
Node Classification Cora CT-Layer Accuracy 67.96% # 71
Node Classification Cornell CT-Layer Accuracy 69.04 # 45
Node Classification Cornell CT-Layer (PE) Accuracy 58.02 # 47
Graph Classification IMDB-BINARY GAP-Layer (Rcut) Accuracy 69.93 # 1
Graph Classification IMDB-BINARY GAP-Layer (Ncut) Accuracy 68.8 # 3
Graph Classification IMDB-BINARY CT-Layer Accuracy 69.84 # 2
Graph Classification MUTAG GAP-Layer (Ncut) Accuracy 86.9% # 48
Graph Classification MUTAG GAP-Layer (Rcut) Accuracy 86.9% # 48
Graph Classification MUTAG CT-Layer Accuracy 87.58% # 42
Graph Classification PROTEINS CT-Layer Accuracy 75.38% # 53
Graph Classification PROTEINS GAP-Layer (Rcut) Accuracy 75.03% # 59
Graph Classification PROTEINS GAP-Layer (Ncut) Accuracy 75.34% # 55
Graph Classification PROTEINS DiffWire Accuracy 74.91% # 61
Node Classification Pubmed CT-Layer Accuracy 68.19 # 64
Node Classification Pubmed CT-Layer (PE) Accuracy 86.07 # 20
Graph Classification REDDIT-BINARY GAP-Layer (Rcut) Accuracy 77.63 # 2
Graph Classification REDDIT-BINARY CT-Layer Accuracy 78.45 # 1
Graph Classification REDDIT-BINARY GAP-Layer (Ncut) Accuracy 76 # 4
Graph Classification REDDIT-BINARY DiffWire Accuracy 77.17 # 3
Node Classification Wisconsin CT-Layer Accuracy 79.05 # 45
Node Classification Wisconsin CT-Layer (PE) Accuracy 69.25 # 48

Methods