Search papers, labs, and topics across Lattice.
University of Michigan
4
0
6
GenRes++ not only detects AI-generated images but also adapts to multiple transformations, outperforming existing methods by focusing on the most informative features.
Achieving 97.6% accuracy in audio deepfake attribution, this framework transforms how we tackle unseen synthesizers in open-set scenarios.
Zero-shot classification accuracy hinges more on the *definition* of a category than the model architecture itself.
Forget retraining: detecting audio deepfakes is as simple as spotting the "seams" in a speech foundation model's embedding trajectory.
