Search papers, labs, and topics across Lattice.
The paper introduces Effective Resistance Rewiring (ERR), a novel graph rewiring strategy that mitigates over-squashing in Graph Neural Networks (GNNs) by optimizing global connectivity. ERR iteratively adds edges between nodes with high effective resistance and removes edges with low resistance, guided by a global signal that aggregates all paths between node pairs. Experiments on various graph types demonstrate that ERR improves signal propagation and predictive performance, particularly when combined with normalization techniques to balance over-squashing and oversmoothing.
By strategically adding edges based on effective resistance, this rewiring method lets GNNs capture long-range dependencies without turning into a homogenous blob.
Graph Neural Networks struggle to capture long-range dependencies due to over-squashing, where information from exponentially growing neighborhoods must pass through a small number of structural bottlenecks. While recent rewiring methods attempt to alleviate this limitation, many rely on local criteria such as curvature, which can overlook global connectivity constraints that restrict information flow. We introduce Effective Resistance Rewiring (ERR), a simple topology correction strategy that uses effective resistance as a global signal to detect structural bottlenecks. ERR iteratively adds edges between node pairs with the largest resistance while removing edges with minimal resistance, strengthening weak communication pathways while controlling graph densification under a fixed edge budget. The procedure is parameter-free beyond the rewiring budget and relies on a single global measure aggregating all paths between node pairs. Beyond predictive performance with GCN models, we analyze how rewiring affects message propagation. By tracking cosine similarity between node embeddings across layers, we examine how the relationship between initial node features and learned representations evolves during message passing, comparing graphs with and without rewiring. This analysis helps determine whether improvements arise from better long-range communication rather than changes in embedding geometry. Experiments on homophilic and heterophilic graphs, including directed settings with DirGCN, reveal a trade-off between over-squashing and oversmoothing, where oversmoothing corresponds to the loss of representation diversity across layers. Resistance-guided rewiring improves connectivity and signal propagation but can accelerate representation mixing in deep models. Combining ERR with normalization techniques such as PairNorm stabilizes this trade-off and improves performance.
