New AI Detector Splits ‘Fake’ Into Subclasses to Boost Generalisation
Researchers propose an SGDFD framework that decomposes AI-generated content detection into semantic and forgery subspaces and splits the single fake ...
Researchers propose an SGDFD framework that decomposes AI-generated content detection into semantic and forgery subspaces and splits the single fake ...
A conceptual five-layer framework proposes that calibrated trust in AI-era photojournalism depends on evidence, traceability, institutional accountability, and audience interpretation ...
A new study shows that feeding wavelet sub-bands into a standard ResNet-50 improves fake-class sensitivity and F1 in cross-forgery deepfake ...
The finding that ear visibility dominates the attribute landscape of DeepFake detection errors deserves closer scrutiny, because it connects to ...
Researchers have developed FedHybrid-ViT, a privacy-preserving framework that fuses convolutional and transformer models within federated learning to detect deepfakes across ...
Researchers have introduced Light-wDEF, a resource-efficient and fully reproducible framework that enabled a fair comparison of ten visual deepfake detectors ...
A new systematic review finds that hybrid deep learning and metaheuristic optimization frameworks achieve 98 to 100 percent accuracy in ...
Researchers developed a lightweight CNN-LSTM model that detects deepfake videos with high accuracy on standard hardware.
A controlled comparison shows attention-enhanced DenseNet121 reaching 97.4 percent accuracy for deepfake detection while a lighter MobileNetV2 variant offers a ...
A new composite scoring framework called RECAF jointly measures deepfake detectors' accuracy, robustness to image distortions, and cross-dataset generalization, revealing ...
© 2026 Scienmag - Science Magazine
© 2026 Scienmag - Science Magazine