HGT-FD: Hypergraph transformer for Fraud Detection

Yintao Cai, Yunjiong Liu, Zelong Yang, Shuyang Fang, Xiaoping Min
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:4996-5004, 2026.

Abstract

Graph-based fraud detection aims to identify anomalous patterns or fraudulent behaviors in graph-structured data, playing a crucial role across various domains. However, traditional models primarily focus on simple node-to-node message passing, which limits the integration of multi-node features and overlooks long-range dependencies in fraud detection. Hyperedge features in hypergraphs, as an integration of node features, can convey and aggregate multi-node characteristics, providing environmental information for the model to detect fraudulent nodes. To this end, we propose HGT-FD for hypergraph-based fraud detection. Specifically, we design a Hypergraph Transformer model that can directly employ hyperedge features for fraud detection, utilizing a co-attention mechanism to generate node representations. Furthermore, structural encoding and positional encoding are proposed to enhance the model’s perception of hypergraph structures, enabling the model to capture more complex high-order structural relationships. Extensive experimental results on three fraud detection datasets demonstrate that the proposed method exhibits significant advantages over baselines in fraud detection.

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-cai26b, title = { HGT-FD: Hypergraph transformer for Fraud Detection }, author = {Cai, Yintao and Liu, Yunjiong and Yang, Zelong and Fang, Shuyang and Min, Xiaoping}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {4996--5004}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/cai26b/cai26b.pdf}, url = {https://proceedings.mlr.press/v300/cai26b.html}, abstract = { Graph-based fraud detection aims to identify anomalous patterns or fraudulent behaviors in graph-structured data, playing a crucial role across various domains. However, traditional models primarily focus on simple node-to-node message passing, which limits the integration of multi-node features and overlooks long-range dependencies in fraud detection. Hyperedge features in hypergraphs, as an integration of node features, can convey and aggregate multi-node characteristics, providing environmental information for the model to detect fraudulent nodes. To this end, we propose HGT-FD for hypergraph-based fraud detection. Specifically, we design a Hypergraph Transformer model that can directly employ hyperedge features for fraud detection, utilizing a co-attention mechanism to generate node representations. Furthermore, structural encoding and positional encoding are proposed to enhance the model’s perception of hypergraph structures, enabling the model to capture more complex high-order structural relationships. Extensive experimental results on three fraud detection datasets demonstrate that the proposed method exhibits significant advantages over baselines in fraud detection. } }
Endnote
%0 Conference Paper %T HGT-FD: Hypergraph transformer for Fraud Detection %A Yintao Cai %A Yunjiong Liu %A Zelong Yang %A Shuyang Fang %A Xiaoping Min %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-cai26b %I PMLR %P 4996--5004 %U https://proceedings.mlr.press/v300/cai26b.html %V 300 %X Graph-based fraud detection aims to identify anomalous patterns or fraudulent behaviors in graph-structured data, playing a crucial role across various domains. However, traditional models primarily focus on simple node-to-node message passing, which limits the integration of multi-node features and overlooks long-range dependencies in fraud detection. Hyperedge features in hypergraphs, as an integration of node features, can convey and aggregate multi-node characteristics, providing environmental information for the model to detect fraudulent nodes. To this end, we propose HGT-FD for hypergraph-based fraud detection. Specifically, we design a Hypergraph Transformer model that can directly employ hyperedge features for fraud detection, utilizing a co-attention mechanism to generate node representations. Furthermore, structural encoding and positional encoding are proposed to enhance the model’s perception of hypergraph structures, enabling the model to capture more complex high-order structural relationships. Extensive experimental results on three fraud detection datasets demonstrate that the proposed method exhibits significant advantages over baselines in fraud detection.
APA
Cai, Y., Liu, Y., Yang, Z., Fang, S. & Min, X.. (2026). HGT-FD: Hypergraph transformer for Fraud Detection . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:4996-5004 Available from https://proceedings.mlr.press/v300/cai26b.html.

Related Material