When Design Rules Break: Benchmark Composition Determines Whether Label Informativeness Predicts GNN Aggregator Choice

• URL: http://arxiv.org/abs/2606.10249v1
• Date: Mon, 08 Jun 2026 23:36:14 GMT
• Title: When Design Rules Break: Benchmark Composition Determines Whether Label Informativeness Predicts GNN Aggregator Choice
• Abstract summary: Graph neural network (GNN) design rules generalize across benchmark families.<n>We study aggregator selection (sum, mean, max) on 24 node-classification datasets spanning citation, heterophilic, LINKX Facebook-100, co-purchase, and co-authorship graphs.<n>Our results suggest that benchmark composition, rather than numerical insufficiency, determines whether design rules appear to generalize.
• Score: 2.1067460775364286
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We examine whether graph neural network (GNN) design rules generalize across benchmark families by studying aggregator selection (sum, mean, max) on 24 node-classification datasets spanning citation, heterophilic, LINKX Facebook-100, co-purchase, and co-authorship graphs. Edge homophily is only weakly predictive of the GIN-Sum versus GIN-Mean performance gap. Label informativeness predicts this gap well on legacy benchmarks but degrades substantially when Facebook-100 graphs are included. In these dense friendship networks, near-zero label informativeness coexists with a strong preference for sum aggregation, producing gains of 7-10% and up to 13% under extended training. Stochastic block model ablations, including degree-corrected variants matching Facebook-100 degree scales, fail to reproduce this behavior, indicating that mean degree alone does not explain the effect. Among several label-independent graph statistics, the spectral gap uniquely distinguishes these graphs from other low-informativeness datasets, with the effect localized to one-hop neighborhoods and replicated across architectures. We further identify training regimes that interact with aggregator choice and show that PNA can underperform the best single-aggregator GIN on standard citation benchmarks. Our results suggest that benchmark composition, rather than numerical insufficiency, determines whether design rules appear to generalize, and that the Facebook-100 regime provides a concrete target for future adaptive aggregation methods.

Related papers

• Subgraph Filtering for Fair Graph Neural Networks [4.075849063907074]
  Graph neural networks (GNNs) can exhibit unfair behavior even when sensitive attributes are excluded from node features.<n>We propose Subgraph Filtering for Fair Graph Neural Networks (SF-GNN)<n>SF-GNN identifies bias-prone edges by combining sensitive homophily with structural propagation amplifiers.<n>It then incorporates edge filtering into each message-passing step to selectively downweight or remove these edges while preserving the remaining graph structure.
  arXiv  Detail & Related papers  (2026-08-26T22:40:31Z)
• PANC: Prior-Aware Normalized Cut for Object Segmentation [0.0]
  We propose a weakly supervised spectral segmentation framework that uses a minimal set of annotated visual tokens.<n>We report strong results on homogeneous, fine-grained, and texture-limited domains.<n>For multi-object benchmarks, the framework showcases explicit, user-controllable semantic segmentation.
  arXiv  Detail & Related papers  (2026-02-06T18:07:20Z)
• DREAM: Dual-Standard Semantic Homogeneity with Dynamic Optimization for Graph Learning with Label Noise [53.55187452152358]
  This paper proposes a novel method, Dual-Standard Semantic Homogeneity with Dynamic Optimization (DREAM) for reliable, relation-informed optimization on graphs with label noise.<n>Specifically, we design a relation-informed dynamic optimization framework that iteratively reevaluates the reliability of each labeled node in the graph.
  arXiv  Detail & Related papers  (2026-01-24T12:54:18Z)
• CGLE: Class-label Graph Link Estimator for Link Prediction [10.111824396356543]
  CGLE (Class-label Graph Link Estimator) is a novel framework designed to augment GNN-based link prediction models.<n>CGLE operates by constructing a class-conditioned link probability matrix.<n>It is encapsulated in an efficient preprocessing stage, leaving the computational complexity of the underlying GNN model unaffected.
  arXiv  Detail & Related papers  (2025-11-10T11:34:36Z)
• Edge Weight Prediction For Category-Agnostic Pose Estimation [12.308036453869033]
  Category-Agnostic Pose Estimation (CAPE) localizes keypoints across diverse object categories with a single model. We introduce EdgeCape, a novel framework that overcomes limitations by predicting the graph's edge weights. We show that this improves the model's ability to capture global spatial dependencies.
  arXiv  Detail & Related papers  (2024-11-25T18:53:09Z)
• The Heterophilic Snowflake Hypothesis: Training and Empowering GNNs for Heterophilic Graphs [59.03660013787925]
  We introduce the Heterophily Snowflake Hypothesis and provide an effective solution to guide and facilitate research on heterophilic graphs. Our observations show that our framework acts as a versatile operator for diverse tasks. It can be integrated into various GNN frameworks, boosting performance in-depth and offering an explainable approach to choosing the optimal network depth.
  arXiv  Detail & Related papers  (2024-06-18T12:16:00Z)
• Chasing Fairness in Graphs: A GNN Architecture Perspective [73.43111851492593]
  We propose textsfFair textsfMessage textsfPassing (FMP) designed within a unified optimization framework for graph neural networks (GNNs) In FMP, the aggregation is first adopted to utilize neighbors' information and then the bias mitigation step explicitly pushes demographic group node presentation centers together. Experiments on node classification tasks demonstrate that the proposed FMP outperforms several baselines in terms of fairness and accuracy on three real-world datasets.
  arXiv  Detail & Related papers  (2023-12-19T18:00:15Z)
• How Graph Structure and Label Dependencies Contribute to Node Classification in a Large Network of Documents [0.0]
  We introduce a new dataset named WikiVitals which contains a large graph of 48k mutually referred Wikipedia articles. Our aim is to rigorously evaluate the contributions of three distinct sources of information to the label prediction in a semi-supervised node classification setting.
  arXiv  Detail & Related papers  (2023-04-03T08:46:43Z)
• Adaptive Graph-Based Feature Normalization for Facial Expression Recognition [1.2246649738388389]
  We propose an Adaptive Graph-based Feature Normalization (AGFN) method to protect Facial Expression Recognition models from data uncertainties. Our method outperforms state-of-the-art works with accuracies of 91.84% and 91.11% on benchmark datasets.
  arXiv  Detail & Related papers  (2022-07-22T14:57:56Z)
• Exploiting Neighbor Effect: Conv-Agnostic GNNs Framework for Graphs with Heterophily [58.76759997223951]
  We propose a new metric based on von Neumann entropy to re-examine the heterophily problem of GNNs. We also propose a Conv-Agnostic GNN framework (CAGNNs) to enhance the performance of most GNNs on heterophily datasets.
  arXiv  Detail & Related papers  (2022-03-19T14:26:43Z)
• Stable Prediction on Graphs with Agnostic Distribution Shift [105.12836224149633]
  Graph neural networks (GNNs) have been shown to be effective on various graph tasks with randomly separated training and testing data. In real applications, however, the distribution of training graph might be different from that of the test one. We propose a novel stable prediction framework for GNNs, which permits both locally and globally stable learning and prediction on graphs.
  arXiv  Detail & Related papers  (2021-10-08T02:45:47Z)
• Graph Classification by Mixture of Diverse Experts [67.33716357951235]
  We present GraphDIVE, a framework leveraging mixture of diverse experts for imbalanced graph classification. With a divide-and-conquer principle, GraphDIVE employs a gating network to partition an imbalanced graph dataset into several subsets. Experiments on real-world imbalanced graph datasets demonstrate the effectiveness of GraphDIVE.
  arXiv  Detail & Related papers  (2021-03-29T14:03:03Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.