arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1806.00804cs.LGcs.CVstat.ML
收藏

NAM:非对抗无监督域映射

NAM: Non-Adversarial Unsupervised Domain Mapping

Facebook AI Research(Facebook人工智能研究院);Tel Aviv University(特拉维夫大学)

Yedid Hoshen, Lior Wolf

摘要:

最近提出了几种方法,用于在没有先前知识的情况下以对应的形式在域之间翻译图像的任务。现有的方法应用对抗学习来确保映射的源域的分布与目标域无法区分,目标域遭受已知的稳定性问题。此外,大多数方法严重依赖于域之间的“循环”关系,这强制执行一对一的映射。在这项工作中,我们引入了一种替代方法:非对抗映射(NAM),它将目标域生成建模任务与跨域映射任务分开。NAM依赖于目标域的预训练生成模型,并将每个源图像与从目标域合成的图像对齐,同时联合优化域映射函数。它有几个关键优势:更高质量和分辨率的图像翻译,更简单和更稳定的训练和可重用的目标模型。大量的实验验证了我们的方法的优点。

英文摘要:

Several methods were recently proposed for the task of translating images between domains without prior knowledge in the form of correspondences. The existing methods apply adversarial learning to ensure that the distribution of the mapped source domain is indistinguishable from the target domain, which suffers from known stability issues. In addition, most methods rely heavily on `cycle' relationships between the domains, which enforce a one-to-one mapping. In this work, we introduce an alternative method: Non-Adversarial Mapping (NAM), which separates the task of target domain generative modeling from the cross-domain mapping task. NAM relies on a pre-trained generative model of the target domain, and aligns each source image with an image synthesized from the target domain, while jointly optimizing the domain mapping function. It has several key advantages: higher quality and resolution image translations, simpler and more stable training and reusable target models. Extensive experiments are presented validating the advantages of our method.

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。

↑