aiqbench analysis
DeltaGNN introduces a scalable Graph Neural Network architecture that mitigates over-smoothing and over-squashing through an efficient information flow control mechanism. By achieving linear computational overhead, this model enables the detection of long-range interactions in large-scale graphs without sacrificing generalizability across diverse topologies. This advancement directly addresses critical bottlenecks in current GNN expressiveness, offering a practical solution for processing complex graph-structured data.
DeltaGNN: Graph Neural Network with Information Flow Control
arXiv:2501.06002v2 Announce Type: replace Abstract: Graph Neural Networks (GNNs) are popular deep learning models designed to process graph-structured data through recursive neighborhood aggregations in the message passing process. When applied to semi-supervised node classification, the message-passing enables GNNs to understand short-range spatial interactions, but also causes them to suffer from over-smoothing and over-squashing. These challenges hinder model expressiveness and prevent the use of deeper models to capture long-range node interactions (LRIs) within the graph. Popular solutions for LRIs detection are either too expensive to process large graphs due to high time complexity or fail to generalize across diverse graph structures. To address these limitations, we propose a mechanism called \emph{information flow control}, which leverages a novel connectivity measure, called \emph{information flow score}, to address over-smoothing and over-squashing with linear computational overhead, supported by theoretical evidence. Building on this mechanism, we introduce DeltaGNN, to the best of our knowledge among the first \textit{scalable} (featuring linear computational and memory complexity overhead) and \textit{generalizable} (capable of effectively handling graphs with diverse homophily, density, and topology) architectures for long-range and short-range interaction detection. We benchmark our model across 10 real-world datasets, including graphs with varying sizes, topologies, densities, and homophilic ratios, showing superior performance with limited computational complexity. The implementation of the proposed methods are publicly available at https://github.com/basiralab/DeltaGNN.
Abstract:Graph Neural Networks (GNNs) are popular deep learning models designed to process graph-structured data through recursive neighborhood aggregations in the message passing process. When applied to semi-supervised node classification, the message-passing enables GNNs to understand short-range spatial interactions, but also causes them to suffer from over-smoothing and over-squashing. These challenges hinder model expressiveness and prevent the use of deeper models to capture long-range node interactions (LRIs) within the graph. Popular solutions for LRIs detection are either too expensive to process large graphs due to high time complexity or fail to generalize across diverse graph structures. To address these limitations, we propose a mechanism called \emph{information flow control}, which leverages a novel connectivity measure, called \emph{information flow score}, to address over-smoothing and over-squashing with linear computational overhead, supported by theoretical evidence. Building on this mechanism, we introduce DeltaGNN, to the best of our knowledge among the first \textit{scalable} (featuring linear computational and memory complexity overhead) and \textit{generalizable} (capable of effectively handling graphs with diverse homophily, density, and topology) architectures for long-range and short-range interaction detection. We benchmark our model across 10 real-world datasets, including graphs with varying sizes, topologies, densities, and homophilic ratios, showing superior performance with limited computational complexity. The implementation of the proposed methods are publicly available at this https URL.
Machine Learning (cs.LG)
arXiv:2501.06002 [cs.LG]
(or arXiv:2501.06002v2 [cs.LG] for this version)
K. Mancini and I. Rekik, "DeltaGNN: Graph Neural Network With Information Flow Control," in IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026
https://doi.org/10.1109/TPAMI.2026.3725398 DOI(s) linking to related resources
Full text under the publisher's open licence · 252 words
Related coverage
INDEQS: Informed Neural controlled Differential EQuationS
arXiv · Oct 2, 2026
Physics-Informed Graph Neural Networks for Robust AC-Optimal Power Flow
arXiv · Jul 30, 2026
GridSFM: A Foundation Model for Solving AC Optimal Power Flow
arXiv · Sep 26, 2026
Local-Global Geometric Insights for Graph Neural Networks via Entropic Curvature
arXiv · Jul 27, 2026