Quantum-Inspired Contextual Learning for Sparse-Ring Fraud Detection in Dynamic Transaction Graphs

arXiv:2607.09704v1 Announce Type: new Abstract: We present an exploratory benchmark and quantum-inspired modeling prototype for fraud screening in dynamic financial transaction graphs. Coordinated fraud may not be visible from individual transactions alone, but may emerge as a multi-period relational pattern. We focus on sparse-ring fraud, a stylized pattern in which a completed directed cycle is distributed across several days, requiring models to integrate evidence across both time and graph s...