自回归图像生成的水印技术
Watermarking Autoregressive Image Generation
AI总结：
本文提出首个针对自回归图像生成模型的令牌级水印方法，通过定制微调流程和同步层解决反向循环一致性缺失问题，实现鲁棒且可验证的水印检测。
摘要：
对生成模型输出进行水印处理已成为追踪其来源的一种有前景的方法。尽管自回归图像生成模型及其被滥用的可能性引起了广泛关注，但此前尚无工作尝试在令牌级别对其输出进行水印处理。在本工作中，我们提出了首个此类方法，通过将语言模型水印技术适配到这一场景。我们识别出一个关键挑战：缺乏反向循环一致性（RCC），即对生成的图像令牌重新进行令牌化会显著改变令牌序列，从而有效擦除水印。为解决这一问题并使我们的方法对常见图像变换、神经压缩和移除攻击具有鲁棒性，我们引入了（i）一个定制的令牌化器-反令牌化器微调流程，以改善RCC，以及（ii）一个互补的水印同步层。正如我们的实验所示，我们的方法能够实现可靠且鲁棒的水印检测，并具有理论上有依据的p值。代码和模型可在 https://github.com/facebookresearch/wmar 获取。
英文摘要：
Watermarking the outputs of generative models has emerged as a promising approach for tracking their provenance. Despite significant interest in autoregressive image generation models and their potential for misuse, no prior work has attempted to watermark their outputs at the token level. In this work, we present the first such approach by adapting language model watermarking techniques to this setting. We identify a key challenge: the lack of reverse cycle-consistency (RCC), wherein re-tokenizing generated image tokens significantly alters the token sequence, effectively erasing the watermark. To address this and to make our method robust to common image transformations, neural compression, and removal attacks, we introduce (i) a custom tokenizer-detokenizer finetuning procedure that improves RCC, and (ii) a complementary watermark synchronization layer. As our experiments demonstrate, our approach enables reliable and robust watermark detection with theoretically grounded p-values. Code and models are available at https://github.com/facebookresearch/wmar.
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。
