Fraud Detection Datasets — Certified Synthetic Registry
Browse certified synthetic fraud detection and AML datasets with SHA-256 fingerprints and Ed25519 certificates. Each entry can be independently audited — no account or trust in CertifiedData required.
About this category
Synthetic fraud detection datasets simulate transaction patterns, anomaly distributions, and label-imbalanced fraud signals without exposing real cardholder, account, or payment data. When certified, the dataset's SHA-256 fingerprint is bound to an Ed25519 signature — enabling auditors, compliance teams, and model validators to independently confirm integrity. Certified entries are suitable for regulated environments requiring defensible data provenance.
Each certified entry in this category has an Ed25519-signed certificate record and a SHA-256 artifact fingerprint. Independent verification is possible without trusting CertifiedData directly.
Common use cases
- Training binary fraud classifiers on imbalanced synthetic transaction data
- AML transaction monitoring model development
- Testing detection thresholds without exposing real fraud cases
- Benchmarking anomaly detection algorithms with labeled ground truth
- Red team simulation and adversarial testing scenarios
Why certification matters
Fraud detection models trained on certified synthetic data can be independently audited — the certificate proves the training set was synthetic and unaltered, reducing liability exposure and satisfying compliance teams who review model lineage.
No entries in this category yet.
Generate and certify a dataset →Register and certify your artifact
Submit an artifact via the API or dashboard. Once certified, it receives an Ed25519-signed certificate and appears in this registry with a live verification link.