Heterogeneous Graph Neural Networks for Malicious Account Detection
This work addresses malicious account detection for mobile payment platforms like Alipay, representing an incremental improvement with a novel method for a known bottleneck.
The paper tackles the problem of detecting malicious accounts on Alipay by introducing GEM, a heterogeneous graph neural network approach that learns embeddings from account-device graphs based on device and activity aggregation weaknesses, achieving promising results compared to competitive methods over time.
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.