PALADIN : Robust Neural Fingerprinting for Text-to-Image Diffusion Models

TL;DR

This paper introduces PALADIN, a neural fingerprinting method for text-to-image diffusion models that achieves near-perfect attribution accuracy, addressing security concerns in open-source generative models.

Contribution

The paper proposes a novel neural fingerprinting technique using cyclic error correcting codes to improve attribution accuracy in diffusion models.

Findings

Achieves near 100% attribution accuracy.

Balances attribution accuracy with generation quality.

Introduces a coding theory-based approach for neural fingerprinting.

Abstract

The risk of misusing text-to-image generative models for malicious uses, especially due to the open-source development of such models, has become a serious concern. As a risk mitigation strategy, attributing generative models with neural fingerprinting is emerging as a popular technique. There has been a plethora of recent work that aim for addressing neural fingerprinting. A trade-off between the attribution accuracy and generation quality of such models has been studied extensively. None of the existing methods yet achieved 100% attribution accuracy. However, any model with less than cent percent accuracy is practically non-deployable. In this work, we propose an accurate method to incorporate neural fingerprinting for text-to-image diffusion models leveraging the concepts of cyclic error correcting codes from the literature of coding theory.