Natural language generation (NLG) applications have gained great popularity due to the powerful deep learning techniques and large training corpus. The deployed NLG models may be stolen or used without authorization, while watermarking has become a useful tool to protect Intellectual Property (IP) of deep models. However, existing watermarking technologies using backdoors are easily detected or harmful for NLG applications. In this paper, we propose a semantic and robust watermarking scheme for NLG models that utilize unharmful phrase pairs as watermarks for IP protection. The watermarks give NLG models personal preference for some special phrase combinations. Specifically, we generate watermarks by following a semantic combination pattern and systematically augment the watermark corpus to enhance the robustness. Then, we embed these watermarks into an NLG model without misleading its original attention mechanism. We conduct extensive experiments and the results demonstrate the effectiveness, robustness, and undetectability of the proposed scheme.
翻译:由于强大的深层学习技巧和大量培训,自然语言生成应用已广受欢迎。部署的自然语言模型可能未经批准被盗或使用,而水标识已成为保护深层模型知识产权的有用工具;然而,利用现有的后门水标识技术很容易检测出来,或对自然语言生成应用有害。在本文件中,我们提议为NLG模型制定一个语义和稳健的水标识计划,利用非有害词组作为知识产权保护的标记。水商标给NLG模型个人偏好某些特殊词组组合。具体地说,我们通过采用语义组合模式并系统地扩大水标识体以加强稳健性,从而生成水标识。然后,我们将这些水标识嵌入国家语言标识模型,而不会误导其原有的注意机制。我们进行了广泛的实验,结果证明了拟议办法的有效性、稳健性和不可探测性。