Heterogeneous Graph Neural Networks for Malicious Account Detection

By Ziqi Liu, Chaochao Chen, Xinxing Yang, Jun Zhou, Xiaolong Li, Le Song. International Conference on Information and Knowledge Management, 2018

415 citations recorded by the source corpus.

We present, GEM, the first heterogeneous graph neural network approach for detecting malicious accounts at Alipay, one of the world's leading mobile cashless payment platform. Our approach, inspired from a connected subgraph approach, adaptively learns discriminative embeddings from heterogeneous account-device graphs based on two fundamental weaknesses of attackers, i.e. device aggregation and activity aggregation. For the heterogeneous graph consists of various types of nodes, we propose an attention mechanism to learn the importance of different types of nodes, while using the sum operator for modeling the aggregation patterns of nodes in each type. Experiments show that our approaches consistently perform promising results compared with competitive methods over time.

Subfields: Artificial Intelligence, Computer Networks and Communications

View Paper: https://www.semanticscholar.org/paper/a5b5dde1da8ac8d9247a32172f76d5d1831309a0

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Philip S. Yu
2. Chuan Shi
3. Xiao Wang
4. Shirui Pan
5. Peng Cui
6. Charu C. Aggarwal
7. Yanfang Ye
8. Chuan Zhou

In Computer Networks and Communications

1. Philip S. Yu
2. Chuan Shi
3. Xiao Wang
4. Peng Cui
5. Ananthram Swami
6. Nitesh V. Chawla
7. Yutong Deng
8. Le Song

Citations in the corpus

• Heterogeneous Graph Neural Network (2019, background): “, GEM [16] for malicious accounts detection, have been proposed to obtain better node embeddings for specific tasks.”
• Anomaly Detection on Attributed Networks via Contrastive Self-Supervised Learning (2021, background): “…attributed graphs), where nodes with attributes indicate real-world entities and links indicate the relationship between entities, are ubiquitous in various scenarios, including finance (trading networks) [1], social media (social networks) [2], [3], and e-commerce (item-user networks) [4], [5].”
• A Semi-Supervised Graph Attentive Network for Financial Fraud Detection (2019, methodology): “Then some following methods start to use graph embedding to incorporate the user interactions [5], [6].”
• Bandit Samplers for Training Graph Neural Networks (2020, background): “For instance, graph neural networks have demonstrated state-of-the-art performance on learning tasks such as node classification, link and graph property prediction, with applications ranging from drug design [8], social networks [11], transaction networks [14], gene expression networks [9], and kno”
• Enhancing Graph Neural Network-based Fraud Detectors against Camouflaged Fraudsters (2020, methodology): “Recently, as the development of Graph Neural Networks (GNNs) (e.g., GCN [17], GAT [34], and GraphSAGE [12]), many GNN-based fraud detectors have been proposed to detect opinion fraud [19, 25, 39], financial fraud [23, 24, 37], mobile fraud [41], and cyber criminal [48].”
• Explainable Artificial Intelligence Applications in Cyber Security: State-of-the-Art in Research (2022, result): “Heterogeneity in transaction graphs975 was captured whereas the presented methodology outper-976 formed previousmodels HGT [191] andGEM [192].”
• Alleviating the Inconsistency Problem of Applying Graph Neural Network to Fraud Detection (2020, background): “Among those frameworks, [8, 15] detect opinion fraud in the online review system, [9, 14, 17] aim at financial fraud, and [16] targets cyber-criminal in online forums.”
• A Survey on Heterogeneous Graph Embedding: Methods, Techniques, Applications and Sources (2020, positive): “[17], [18], many of which [6], [19], [20], [21], [22], [23] have demonstrated the success of heterogeneous graph embedding techniques deployed in real-world applications including recommendation systems [2], [3], malware detection systems [7], [22], [23], [24], and healthcare systems [25], [26].”
• A Comprehensive Survey on Graph Anomaly Detection With Deep Learning (2021, neutral): “com/raghavchalapathy/gad GEM [134] Python Static Graph https://github.”
• Pick and Choose: A GNN-based Imbalanced Learning Approach for Fraud Detection (2021, background): “GEM [25] adaptively learns discriminative embeddings from heterogeneous account-device graphs for malicious accounts detection.”