#1 out of 7562.0 est. views91.30%
technology4h ago
Artificial Intelligence Reveals Fingerprints Aren't Truly Unique, Debunking 100 Years of Forensic Science
- AI-trained models found cross-finger similarity across a person’s fingerprints, challenging the idea of universal uniqueness.
- Researchers found ridge orientation and curvature, not minutiae, most influenced cross-finger similarity.
- The AI may speed up investigations by narrowing large suspect pools to a smaller set of likely candidates.
- No courtroom use yet; experts say current accuracy falls below traditional same-finger matching systems.
- Cross-finger similarity remained significant across datasets and even across hands.
- The study tested diverse datasets including NIST SD300, SD302, and RidgeBase.
- Synthetic training data (PrintsGAN) helped pre-train the model before real-world fine-tuning.
- Researchers stress need for diverse, representative data to avoid algorithmic bias.
- The study suggests potential security implications for biometrics beyond forensics.
- Experts caution that current results are not a final verdict on fingerprint uniqueness.
Vote 0






