Formal Definition of Fingerprints Improves Attribution of Generative Models

NeurIPS 2023 Workshop ATTRIB Submission24 Authors

Published: 27 Oct 2023, Last Modified: 08 Dec 2023ATTRIB PosterEveryoneRevisionsBibTeX
Keywords: generative models, fingerprints, model attribution, source identification, artifacts of generative models
TL;DR: We propose formal definitions of artifacts and fingerprints of generative models, and use them to study which choices in the design of generative models determine their fingerprints.
Abstract: Recent works have shown that generative models leave traces of their underlying generative process on the generated samples, broadly referred to as fingerprints of a generative model, and have studied their utility in detecting synthetic images from real ones. However, the extent to which these fingerprints can distinguish between various types of synthetic images and help identify the underlying generative process remain under-explored. In particular, the very definition of a fingerprint remains unclear, to our knowledge. To that end, in this work, we formalize the definition of artifact and fingerprint in generative models, propose an algorithm for computing them in practice, and finally study how different design parameters affect the model fingerprints and their attributability. We find that using our proposed definition can significantly improve the performance on the task of identifying the underlying generative process from samples (model attribution) compared to existing methods. Additionally, we study the structure of the fingerprints and observe that it is very predictive of the effect of different design choices on the generative process.
Submission Number: 24