图神经网络论文趋势 2023–2026

主要结论

• 图神经网络论文从 2023 年的 434 篇降到 2026 年的 183 篇，热度在回落。
• NeurIPS 是发图神经网络论文最多的会议，2023–2026 年共 489 篇，占四个会议总量的 47%。
• 交叉方向里涨得最快的是数据集与基准：同时属于两个主题的论文从 2023 年的 10 篇增至 2026 年的 10 篇。
• 在「图与结构化数据」的 5 个主题里，它的 2026 年论文数排第 4；同组增长最快的时间序列是 3.2 倍，图神经网络是 0.4 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026A Differentiable 3D Scene Graph Metric via Contextual Hellinger Triplet Geometry
• ICML 2026Aitchison Embeddings for Learning Compositional Graph Representations
• ICLR 2026Certified Evaluation of Model-Level Explanations for Graph Neural Networks
• CVPR 2026Gamba Mamba-based graph convolutional network with dynamic graph topology learning
• NeurIPS 2025Copresheaf Topological Neural Networks: A Generalized Deep Learning Framework
• ICML 2025Disentangling Invariant Subgraph via Variance Contrastive Estimation under Distribution Shifts
• ICLR 2025Edge Prompt Tuning for Graph Neural Networks
• CVPR 2025Coeff-Tuning: A Graph Filter Subspace View for Tuning Attention-Based Large Models
• NeurIPS 2024[Re] Reproducibility Study of “Explaining Temporal Graph Models Through an Explorer-Navigator Framework"
• ICML 2024A Graph is Worth $K$ Words: Euclideanizing Graph using Pure Transformer
• ICLR 2024Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness
• CVPR 2024FlowerFormer: Empowering Neural Architecture Encoding using a Flow-aware Graph Transformer
• NeurIPS 2023[Re] On Explainability of Graph Neural Networks via Subgraph Explorations
• ICML 2023Beyond Homophily: Reconstructing Structure for Graph-agnostic Clustering
• ICLR 2023$\mathscr{N}$-WL: A New Hierarchy of Expressivity for Graph Neural Networks
• CVPR 2023Adversarially Robust Neural Architecture Search for Graph Neural Networks
• NeurIPS 2026A Topological Encoder Decoder Framework for Temporal Graph Learning
• ICML 2026Any-dimensional invariant universality
• ICLR 2026Laplacian Kernelized Bandit
• CVPR 2026Global-Aware Edge Prioritization for Pose Graph Initialization

全部图神经网络论文

共 1,038 篇，第 2 / 21 页

51. ICML 2026Boundary Embedding Shaping with Adaptive Contrastive Learning for Graph Structural Disentanglement
52. ICML 2026Brain Networks Should Be Learned, Not Constructed
53. ICML 2026CAffNet: Hard Constraint-Affine Neural Networks
54. ICML 2026Capacity-Agnostic Parameter Isolation for Continual Graph Learning
55. ICML 2026CLINIC: Towards High-quality Graph Out-Of-Distribution Detection
56. ICML 2026Controlled Dynamics Attractor Transformer
57. ICML 2026Delving into Non-Exchangeability for Conformal Prediction in Graph-Structured Multivariate Time Series
58. ICML 2026Discriminative Attribute Graph Clustering Through Topology-Guided Contrastive Learning
59. ICML 2026Efficient Code Analysis via Graph Representation Learning-Guided Large Language Models
60. ICML 2026Efficiently Training Time-to-First-Spike Spiking Neural Networks from Scratch
61. ICML 2026EpiTwin: Spatiotemporal Graph Transformers for Epileptic sEEG Signal Reconstruction
62. ICML 2026Error-Driven Graph Augmentation for Mesh-Based PDE Surrogates
63. ICML 2026Explicitly Modeling Censoring Produces Superior Survival Predictors
64. ICML 2026Expressive Graph Neural Networks via Equivariant Use of Noise
65. ICML 2026FedRGL: Robust Federated Graph Learning under Label Noise
66. ICML 2026From Distribution to Geometry: Stable Graph Generalization via Invariant Barycenters
67. ICML 2026Full-Spectrum Graph Neural Networks: Expressive and Scalable
68. ICML 2026Gauge-Equivariant Graph Networks via Self-Interference Cancellation
69. ICML 2026GI-GCN: Global Interacted Graph Convolutional Networks via Dominant Sets for Graph Classification
70. ICML 2026Graph is a Natural Regularization: Revisiting Vector Quantization for Graph Representation Learning
71. ICML 2026Graph is a Substrate Across Data Modalities
72. ICML 2026Graph Neural Networks Are Not Continuous Across Graph Resolutions
73. ICML 2026GraphP-FL: Personalized Federated Graph Learning via Dynamic Structure Awareness and Fisher Information Elastic Alignment
74. ICML 2026InvGNN: Learning Invertible Node Representations on Graphs
75. ICML 2026Is Fixing Schema Graphs Necessary? Full-Resolution Graph Structure Learning for Relational Deep Learning
76. ICML 2026L2G-NET: Local to Global Spectral Graph Neural Networks via Cauchy Factorizations
77. ICML 2026Learning Molecular Semantic Invariant Representation with Prototype Constraint
78. ICML 2026Learning to Approximate Uniform Facility Location via Graph Neural Networks
79. ICML 2026Learning to Execute Graph Algorithms Exactly with Graph Neural Networks
80. ICML 2026NetDiff: Graph Diffusion with Improved Global Capabilities to Generate and Update Mobile Network Topologies
81. ICML 2026On the Expressive Power of GNNs to Solve Linear SDPs
82. ICML 2026On the Spectral Unreachability of Brain Graph Learning
83. ICML 2026One-Step Graph-Structured Neural Flows for Irregular Multivariate Time Series Classification
84. ICML 2026OpenMAG: A Comprehensive Benchmark for Multimodal-Attributed Graph
85. ICML 2026Physics-informed coarsening for multigrid graph neural networks surrogates
86. ICML 2026Quantile-Free Uncertainty Quantification in Graph Neural Networks
87. ICML 2026RAPNet: Accelerating Algebraic Multigrid with Learned Sparse Corrections
88. ICML 2026Rel-MOSS: Towards Imbalanced Relational Deep Learning on Relational Databases
89. ICML 2026Rethinking Graph Transformers as Graph Signal Denoisers: The Role of Block-Diagonal Priors
90. ICML 2026Rethinking Parameter Sharing as Graph Coloring for Structured Compression
91. ICML 2026Revisiting Positive Samples in Graph Contrastive Learning: From the Perspective of Message Passing
92. ICML 2026S$^3$GNN: Efficient Global Mixing and Local Message Passing for Long-Range Graph Learning
93. ICML 2026SAGE-NAS: Synergizing LLM-Based Semantic Agent with Graph-Based Evaluator for Neural Architecture Search
94. ICML 2026SC-FAGC: Size Constrained Fast Anchor-based Graph Clustering
95. ICML 2026SFCLTA: Spectral Fusion Contrastive Learning with Topology-Adaptive Graph Augmentation
96. ICML 2026Sheaf Neural Networks on SPD Manifolds: Second-Order Geometric Representation Learning
97. ICML 2026Sparse Autoencoders are Topic Models
98. ICML 2026Spike-HTR: Spiking Neural Transformer for Handwritten Text Recognition
99. ICML 2026SWING: Unlocking Implicit Graph Representations for Graph Random Features
100. ICML 2026Target-Oriented Pretraining Data Selection via Neuron-Activated Graph

常见问题

2023–2026 年四大会议一共有多少篇图神经网络论文？

共 1,038 篇，按年份分别是 2023 年 434 篇、2024 年 208 篇、2025 年 213 篇、2026 年 183 篇。

NeurIPS 2026 有多少篇图神经网络论文？

NeurIPS 2026 有 74 篇图神经网络论文，上一届是 75 篇。

ICLR 2026 有多少篇图神经网络论文？

ICLR 2026 有 28 篇图神经网络论文，上一届是 45 篇。

ICML 2026 有多少篇图神经网络论文？

ICML 2026 有 68 篇图神经网络论文，上一届是 81 篇。

CVPR 2026 有多少篇图神经网络论文？

CVPR 2026 有 13 篇图神经网络论文，上一届是 12 篇。

哪个会议的图神经网络论文最多？

NeurIPS，2023–2026 年共 489 篇。