Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
144 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Provably Powerful Graph Neural Networks for Directed Multigraphs (2306.11586v3)

Published 20 Jun 2023 in cs.LG and cs.AI

Abstract: This paper analyses a set of simple adaptations that transform standard message-passing Graph Neural Networks (GNN) into provably powerful directed multigraph neural networks. The adaptations include multigraph port numbering, ego IDs, and reverse message passing. We prove that the combination of these theoretically enables the detection of any directed subgraph pattern. To validate the effectiveness of our proposed adaptations in practice, we conduct experiments on synthetic subgraph detection tasks, which demonstrate outstanding performance with almost perfect results. Moreover, we apply our proposed adaptations to two financial crime analysis tasks. We observe dramatic improvements in detecting money laundering transactions, improving the minority-class F1 score of a standard message-passing GNN by up to 30%, and closely matching or outperforming tree-based and GNN baselines. Similarly impressive results are observed on a real-world phishing detection dataset, boosting three standard GNNs' F1 scores by around 15% and outperforming all baselines.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (73)
  1. The surprising power of graph neural networks with random node initialization. arXiv preprint arXiv:2010.01179.
  2. Realistic Synthetic Financial Transactions for Anti-Money Laundering Models. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track.
  3. Random graph isomorphism. SIaM Journal on computing, 9(3): 628–635.
  4. Graph neural networks with local graph parameters. Advances in Neural Information Processing Systems, 34: 25280–25293.
  5. Interaction networks for learning about objects, relations and physics. Advances in neural information processing systems, 29.
  6. Relational inductive biases, deep learning, and graph networks. arXiv preprint arXiv:1806.01261.
  7. Scalable Fine-Grained Parallel Cycle Enumeration Algorithms. In Proceedings of the 34th ACM Symposium on Parallelism in Algorithms and Architectures, 247–258.
  8. Fast Parallel Algorithms for Enumeration of Simple, Temporal, and Hop-Constrained Cycles. ArXiv:2301.01068 [cs].
  9. Molecular generative graph neural networks for drug discovery. Neurocomputing, 450: 242–252.
  10. Improving graph neural network expressivity via subgraph isomorphism counting. IEEE Transactions on Pattern Analysis and Machine Intelligence.
  11. LaundroGraph: Self-Supervised Graph Representation Learning for Anti-Money Laundering. In Proceedings of the Third ACM International Conference on AI in Finance, 130–138.
  12. XBLOCK Blockchain Datasets: InPlusLab Ethereum Phishing Detection Datasets. http://xblock.pro/ethereum/.
  13. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, 785–794.
  14. Incorporating corporation relationship via graph convolutional neural networks for stock price prediction. In Proceedings of the 27th ACM International Conference on Information and Knowledge Management, 1655–1658.
  15. Can graph neural networks count substructures? Advances in neural information processing systems, 33: 10383–10395.
  16. On the equivalence between graph isomorphism testing and function approximation with gnns. Advances in neural information processing systems, 32.
  17. Eta prediction with graph neural networks in google maps. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management, 3767–3776.
  18. Graph neural networks with learnable structural and positional representations. arXiv preprint arXiv:2110.07875.
  19. Graph Neural Networks with Precomputed Node Features. arXiv preprint arXiv:2206.00637.
  20. Temporal relational ranking for stock prediction. ACM Transactions on Information Systems (TOIS), 37(2): 1–30.
  21. Understanding and extending subgraph gnns by rethinking their symmetries. arXiv preprint arXiv:2206.11140.
  22. Generalization and representational limits of graph neural networks. In International Conference on Machine Learning, 3419–3430. PMLR.
  23. The geometry of suspicious money laundering activities in financial networks. EPJ Data Science, 11(1): 6.
  24. Inductive representation learning on large graphs. Advances in neural information processing systems, 30.
  25. An efficient solution to detect common topologies in money launderings based on coupling and connection. IEEE Intelligent Systems, 36(1): 64–74.
  26. Multilayer feedforward networks are universal approximators. Neural networks, 2(5): 359–366.
  27. Open graph benchmark: Datasets for machine learning on graphs. Advances in neural information processing systems, 33: 22118–22133.
  28. Strategies for pre-training graph neural networks. arXiv preprint arXiv:1905.12265.
  29. Boosting the Cycle Counting Power of Graph Neural Networks with I22{}^{2}start_FLOATSUPERSCRIPT 2 end_FLOATSUPERSCRIPT-GNNs. arXiv preprint arXiv:2210.13978.
  30. edGNN: a Simple and Powerful GNN for Directed Labeled Graphs. arXiv preprint arXiv:1904.08745.
  31. Ethereum Fraud Detection with Heterogeneous Graph Neural Networks. arXiv preprint arXiv:2203.12363.
  32. Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems, 30.
  33. Keisler, R. 2022. Forecasting global weather with graph neural networks. arXiv preprint arXiv:2202.07575.
  34. Semi-Supervised Classification with Graph Convolutional Networks. In International Conference on Learning Representations.
  35. Hierarchical Representation Learning for Bipartite Graphs. In IJCAI, volume 19, 2873–2879.
  36. Modeling the stock relation with graph network for overnight stock movement prediction. In Proceedings of the twenty-ninth international conference on international joint conferences on artificial intelligence, 4541–4547.
  37. Uncovering insurance fraud conspiracy with network learning. In Proceedings of the 42nd international ACM SIGIR conference on research and development in information retrieval, 1181–1184.
  38. Large scale learning on non-homophilous graphs: New benchmarks and strong simple methods. Advances in Neural Information Processing Systems, 34: 20887–20902.
  39. Heterogeneous graph neural networks for malicious account detection. In Proceedings of the 27th ACM international conference on information and knowledge management, 2077–2085.
  40. Inspection-L: Practical GNN-based money laundering detection system for bitcoin. arXiv preprint arXiv:2203.10465.
  41. Loukas, A. 2019. What graph neural networks cannot learn: depth vs width. arXiv preprint arXiv:1907.03199.
  42. Spectral-based graph convolutional network for directed graphs. arXiv preprint arXiv:1907.08990.
  43. Provably powerful graph networks. Advances in neural information processing systems, 32.
  44. Medvedev, E. 2022. ETH-etl. https://github.com/blockchain-etl/ethereum-etl. Accessed: 01-12-2022.
  45. Weisfeiler and leman go neural: Higher-order graph neural networks. In Proceedings of the AAAI conference on artificial intelligence, volume 33, 4602–4609.
  46. Financial cybercrime: A comprehensive survey of deep learning approaches to tackle the evolving financial crime landscape. Ieee Access, 9: 163965–163986.
  47. DropGNN: Random dropouts increase the expressiveness of graph neural networks. Advances in Neural Information Processing Systems, 34: 21997–22009.
  48. A theoretical comparison of graph neural network extensions. In International Conference on Machine Learning, 17323–17345. PMLR.
  49. Geom-gcn: Geometric graph convolutional networks. arXiv preprint arXiv:2002.05287.
  50. xFraud: explainable fraud transaction detection. Proceedings of the VLDB Endowment, 15: 427–436.
  51. Gradient gating for deep multi-rate learning on graphs. arXiv preprint arXiv:2210.00513.
  52. Approximation ratios of graph neural networks for combinatorial problems. Advances in Neural Information Processing Systems, 32.
  53. Random features strengthen graph neural networks. In Proceedings of the 2021 SIAM International Conference on Data Mining (SDM), 333–341. SIAM.
  54. Modeling relational data with graph convolutional networks. In Extended Semantic Web Conference, 593–607. Springer.
  55. Beyond news contents: The role of social context for fake news detection. In Proceedings of the twelfth ACM international conference on web search and data mining, 312–320.
  56. Smurf-Based Anti-money Laundering in Time-Evolving Transaction Networks. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, 171–186. Springer.
  57. Suzumura, T. 2022. AMLSIM library wiki. https://github.com/IBM/AMLSim/wiki/Transaction-Model:-Alert-Model. Accessed: 30-11-2022.
  58. Digraph inception convolutional networks. Advances in neural information processing systems, 33: 17907–17918.
  59. Graph attention networks. arXiv preprint arXiv:1710.10903.
  60. Deep Graph Infomax. ICLR (Poster), 2(3): 4.
  61. Collective dynamics of ‘small-world’networks. nature, 393(6684): 440–442.
  62. Scalable graph learning for anti-money laundering: A first look. arXiv preprint arXiv:1812.00076.
  63. Anti-money laundering in bitcoin: Experimenting with graph convolutional networks for financial forensics. arXiv preprint arXiv:1908.02591.
  64. A comprehensive survey on graph neural networks. IEEE transactions on neural networks and learning systems, 32(1): 4–24.
  65. Towards consumer loan fraud detection: Graph neural networks with role-constrained conditional random field. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, 4537–4545.
  66. How powerful are graph neural networks? arXiv preprint arXiv:1810.00826.
  67. Financial risk analysis for SMEs with graph-based supply chain mining. In Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence, 4661–4667.
  68. Identity-aware graph neural networks. In Proceedings of the AAAI conference on artificial intelligence, volume 35, 10737–10745.
  69. Nested graph neural networks. Advances in Neural Information Processing Systems, 34: 15734–15747.
  70. Deep learning based recommender system: A survey and new perspectives. ACM computing surveys (CSUR), 52(1): 1–38.
  71. Magnet: A neural network for directed graphs. Advances in neural information processing systems, 34: 27003–27015.
  72. From stars to subgraphs: Uplifting any GNN with local structure awareness. arXiv preprint arXiv:2110.03753.
  73. Graph neural networks: A review of methods and applications. AI open, 1: 57–81.
Citations (9)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now