Graph neural networks meet the edge in new fraud detection survey
A systematic survey of 22 studies finds that no published fraud-detection system simultaneously combines graph learning, edge deployment, federated training, ...
A systematic survey of 22 studies finds that no published fraud-detection system simultaneously combines graph learning, edge deployment, federated training, ...
A new open-access framework combines graph neural networks, LSTM models, large language models and federated learning to detect money laundering ...
Researchers have unveiled an artificial intelligence system that classifies international trade transactions as low, medium, or high risk with reported ...
A new open-access study reports that a layered fraud detection framework combining verified explanations, probability calibration and cost-sensitive thresholds outperformed ...
A five-stage aggregation framework called Hybrid Fair reports top F1 scores in six of eight model–dataset combinations while cutting blockchain ...
Researchers built a synthetic, Japan-calibrated medical claims benchmark with traceable fraud counterfactuals and a frozen single-access evaluation, reporting strong anti-fraud ...
Researchers report a unified Android fraud detection framework combining kernel-level traffic decryption, a behavioral graph, and time-decay risk scoring that ...
INSIGHT combines mixed-distance clustering, cluster-aware oversampling and interpretability analyses to improve fraud detection in two financial transaction datasets.
Researchers have developed VoteChain, a voting-based blockchain consensus framework that embeds a lightweight CNN-LSTM anomaly detector to cut energy use ...
Researchers have developed CL3AN-GNN, a graph neural network that borrows curriculum learning from education to progressively disentangle features and boost ...
© 2026 Scienmag - Science Magazine
© 2026 Scienmag - Science Magazine