arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.16889cs.LGcs.AIcs.CR
收藏

生成模型是自水印的:通过再生成声明模型认证

Generative Models are Self-Watermarked: Declaring Model Authentication through Re-Generation

University of Melbourne(墨尔本大学);University College London(伦敦大学学院);Macquarie University(麦考瑞大学);Singapore University of Technology and Design(新加坡科技设计大学)

Aditya Desu, Xuanli He, Qiongkai Xu, Wei Lu

AI总结:

本文提出通过再生成检测生成模型输出中天然存在的潜在指纹,实现数据所有权归属验证,并通过迭代再生成放大指纹,在文本和图像生成模型上验证了方法的可行性与鲁棒性。

摘要:

随着机器和人工智能生成内容的激增,保护生成模型的知识产权已变得势在必行,然而验证数据所有权构成了巨大挑战,尤其是在未经授权重用生成数据的情况下。使用机器学习即服务(MLaaS)进一步放大了验证数据所有权的挑战,因为MLaaS通常以黑盒系统的形式运行。我们的工作致力于检测哪怕单个样本的数据重用。传统上,水印技术被用于检测人工智能生成的内容。然而,与水印技术将额外信息作为触发器嵌入模型或生成内容中、可能损害输出质量不同,我们的方法通过再生成来识别输出中天然存在的潜在指纹。我们提出了一种可解释的验证程序,通过再生成来归属数据所有权,并进一步通过迭代数据再生成在生成模型中放大这些指纹。该方法具有理论基础,并使用近期先进的文本和图像生成模型证明了其可行性和鲁棒性。我们的方法意义重大,因为它超越了保护API知识产权的范畴,并解决了诸如错误信息传播和学术不端等重要问题。它提供了一种有用的工具来确保来源和作者身份的完整性,将其应用扩展到真实性和所有权验证至关重要的不同场景中。

英文摘要:

As machine- and AI-generated content proliferates, protecting the intellectual property of generative models has become imperative, yet verifying data ownership poses formidable challenges, particularly in cases of unauthorized reuse of generated data. The challenge of verifying data ownership is further amplified by using Machine Learning as a Service (MLaaS), which often functions as a black-box system. Our work is dedicated to detecting data reuse from even an individual sample. Traditionally, watermarking has been leveraged to detect AI-generated content. However, unlike watermarking techniques that embed additional information as triggers into models or generated content, potentially compromising output quality, our approach identifies latent fingerprints inherently present within the outputs through re-generation. We propose an explainable verification procedure that attributes data ownership through re-generation, and further amplifies these fingerprints in the generative models through iterative data re-generation. This methodology is theoretically grounded and demonstrates viability and robustness using recent advanced text and image generative models. Our methodology is significant as it goes beyond protecting the intellectual property of APIs and addresses important issues such as the spread of misinformation and academic misconduct. It provides a useful tool to ensure the integrity of sources and authorship, expanding its application in different scenarios where authenticity and ownership verification are essential.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑