Quantum Assistant Detects Rare Frauds ⚡ экспресс
Credit fraud detection suffers from class imbalance, causing models to miss fraudulent transactions. Q-SYNTH—a hybrid GAN with a quantum generator and classical discriminator—is proposed. The method was evaluated on statistical similarity (KS statistic, Wasserstein distance) and downstream classification. Compared to classical GAN, Q-SYNTH reduces marginal distribution divergence while preserving detection quality. Although SMOTE better reproduces individual features, and classical GAN achieves higher metrics under some conditions, the quantum approach offers an optimal compromise between fidelity and practical utility. This confirms the applicability of hybrid quantum augmentation in fraud detection tasks.
In the ocean of bank transactions, fraud is rarer than neutron stars in space—just 0.1% of all operations. A conventional detector gets used to the clean flow and loses vigilance.
Enter Q-SYNTH, a hybrid system with a quantum generator. It creates plausible fake transactions itself, drawing on scraps of real data. The quantum mechanism, inspired by the ideas of Alexei Kitaev, works like a supernova: from a small number of parameters, it unfolds a whole spectrum of new examples. These synthetic data are so high-quality that even the built-in “discriminator” sometimes can’t tell them from real ones.
Thanks to this, the detector learns to see hidden patterns. The artificial examples reduce entropy—a measure of chaos in the data—helping to pick up signals, akin to gravitational waves from invisible cataclysms. As a result, Q-SYNTH finds the sweet spot: it misses fewer real scams without overloading the bank with false alarms.
🎯 Only 0.1% of bank transactions are fraudulent. Without training on fakes, a detector almost always says 'clean' and misses rare but devastating scams.
🎬 The technology resembles a lie detector from the future, learning from fictional crimes to catch real ones.