Benchmarking Sheaf Neural Networks for Inductive Tasks

• URL: http://arxiv.org/abs/2608.02558v1
• Date: Mon, 03 Aug 2026 17:41:20 GMT
• Title: Benchmarking Sheaf Neural Networks for Inductive Tasks
• Abstract summary: Sheaf Networks (SNN) generalize message passing by replacing scalar weights of standard Graph Networks (GNNs) with learnable, edge-dependent restriction stalks.<n>We show that SNNs transfer to inductive settings but do not reach the strongest baselines.<n> Practically, a single sheaf configuration can generalize across datasets so effort is better spent tuning the surrounding architectural recipe.
• Score: 14.028547763374425
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Sheaf Neural Networks (SNNs) generalize message passing by replacing scalar edge weights of standard Graph Neural Networks (GNNs) with learnable, edge-dependent restriction maps between node stalks. Despite their strong theoretical foundations and promising transductive results, SNNs have been evaluated almost exclusively on transductive node classification, leaving their behaviour under inductive protocols unknown. We address this gap through the first systematic benchmark of the sheaf design space, evaluating three diffusion mechanisms (neural sheaf diffusion, sheaf attention, and sheaf attention with Graph Attention Network v2), three restriction-map parameterizations, three stalk dimensions, and six modern GNN architectural components, within a message-passing reformulation that never assembles the heavy sheaf Laplacian, making the full design space trainable under cross-graph batching. Across $1{,}890$ controlled experiments on 14 inductive datasets, multiple insights emerge: restriction maps are the dominant design choice and general maps are preferable, larger stalks add capacity but not long-range reach, architectural components explain more performance variation than the entire sheaf-specific design space itself. Under a matched protocol, SNNs transfer to inductive settings but do not reach the strongest baselines, with gaps being dataset-dependent. Practically, a single sheaf configuration can generalize across datasets, so effort is better spent tuning the surrounding architectural recipe than the sheaf operator itself.

Related papers

• Differentiable Lifting for Topological Neural Networks [49.208571794040374]
  Topological neural networks (TNNs) enable leveraging high-order structures on graphs.<n>DiffLift is a framework for learning graph liftings to hypergraphs and cellular- and simplicial complexes.<n>Our approach leads to gains of up to 45% over static liftings, including both connectivity- and feature-based ones.
  arXiv  Detail & Related papers  (2026-08-02T11:30:09Z)
• Hierarchical Pooling for Sheaf Neural Networks [20.79323358337401]
  We introduce Hierarchical Sheaf Pool (HiSP), a sheaf-aware pooling framework based on local spectral coarsening.<n>HiSP constructs each coarse stalk by projecting fine stalk-valued features onto the low-frequency eigenmodes of the cluster-internal sheaf Laplacian.<n>We analyze the approximation induced by coarsening by separating truncation loss, due to discarded local modes, from realization loss due to representing the projected operator as coarse sheaf.
  arXiv  Detail & Related papers  (2026-06-18T20:50:27Z)
• On the Necessity of Learnable Sheaf Laplacians [12.949900291633943]
  Sheaf Neural Networks (SNNs) were introduced as an extension of Graph Convolutional Networks to address oversmoothing on heterophilous graphs.<n>We show that Identity Sheaf Networks do not appear to suffer more significant oversmoothing than their SNN counterparts.
  arXiv  Detail & Related papers  (2026-03-05T17:24:13Z)
• Sheaves Reloaded: A Directional Awakening [31.634923352992324]
  Sheaf Neural Networks (SNNs) represent a powerful generalization of Graph Neural Networks (GNNs)<n>We introduce the Directed Cellular Sheaf, a special type of cellular sheaf designed to explicitly account for edge orientation.<n>Building on this structure, we define a new sheaf Laplacian, which captures both the graph's topology and its directional information.<n>This operator serves as the backbone of the Directed Sheaf Neural Network (DSNN), the first SNN model to embed a directional bias into its architecture.
  arXiv  Detail & Related papers  (2025-06-03T13:13:56Z)
• Joint Diffusion Processes as an Inductive Bias in Sheaf Neural Networks [14.224234978509026]
  Sheaf Neural Networks (SNNs) naturally extend Graph Neural Networks (GNNs) We propose two novel sheaf learning approaches that provide a more intuitive understanding of the involved structure maps. In our evaluation, we show the limitations of the real-world benchmarks used so far on SNNs.
  arXiv  Detail & Related papers  (2024-07-30T07:17:46Z)
• Binary Graph Convolutional Network with Capacity Exploration [58.99478502486377]
  We propose a Binary Graph Convolutional Network (Bi-GCN), which binarizes both the network parameters and input node attributes. Our Bi-GCN can reduce the memory consumption by an average of 31x for both the network parameters and input data, and accelerate the inference speed by an average of 51x.
  arXiv  Detail & Related papers  (2022-10-24T12:05:17Z)
• Simple and Efficient Heterogeneous Graph Neural Network [55.56564522532328]
  Heterogeneous graph neural networks (HGNNs) have powerful capability to embed rich structural and semantic information of a heterogeneous graph into node representations. Existing HGNNs inherit many mechanisms from graph neural networks (GNNs) over homogeneous graphs, especially the attention mechanism and the multi-layer structure. This paper conducts an in-depth and detailed study of these mechanisms and proposes Simple and Efficient Heterogeneous Graph Neural Network (SeHGNN)
  arXiv  Detail & Related papers  (2022-07-06T10:01:46Z)
• Deep Architecture Connectivity Matters for Its Convergence: A Fine-Grained Analysis [94.64007376939735]
  We theoretically characterize the impact of connectivity patterns on the convergence of deep neural networks (DNNs) under gradient descent training. We show that by a simple filtration on "unpromising" connectivity patterns, we can trim down the number of models to evaluate.
  arXiv  Detail & Related papers  (2022-05-11T17:43:54Z)
• Dual-constrained Deep Semi-Supervised Coupled Factorization Network with Enriched Prior [80.5637175255349]
  We propose a new enriched prior based Dual-constrained Deep Semi-Supervised Coupled Factorization Network, called DS2CF-Net. To ex-tract hidden deep features, DS2CF-Net is modeled as a deep-structure and geometrical structure-constrained neural network. Our network can obtain state-of-the-art performance for representation learning and clustering.
  arXiv  Detail & Related papers  (2020-09-08T13:10:21Z)
• The Curious Case of Convex Neural Networks [12.56278477726461]
  We show that the convexity constraints can be enforced on both fully connected and convolutional layers. We draw three valuable insights: (a) Input Output Convex Neural Networks (IOC-NNs) self regularize and reduce the problem of overfitting; (b) Although heavily constrained, they outperform the base multi layer perceptrons and achieve similar performance as compared to base convolutional architectures.
  arXiv  Detail & Related papers  (2020-06-09T08:16:38Z)
• Node Masking: Making Graph Neural Networks Generalize and Scale Better [71.51292866945471]
  Graph Neural Networks (GNNs) have received a lot of interest in the recent times. In this paper, we utilize some theoretical tools to better visualize the operations performed by state of the art spatial GNNs. We introduce a simple concept, Node Masking, that allows them to generalize and scale better.
  arXiv  Detail & Related papers  (2020-01-17T06:26:40Z)

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.