记一篇DeepFake主动防御方向的review

hosheaw

于 2022-10-26 18:27:32 发布

阅读量672

点赞数

CC 4.0 BY-SA版权

文章标签：机器学习

本文链接：https://blog.csdn.net/qq_37252651/article/details/127537941

本文提出了一种在多媒体文件发布到公共网络前添加扰动的主动DeepFake防御策略，该策略在实验中表现出良好的防御效果且几乎不引入可见干扰。文章指出了现有DeepFake检测器（尤其是被动检测器）的重要问题，并在HSV空间添加扰动方面具有一定创新。然而，论文的创新性总体不足，缺乏对增强鲁棒性的解释，使用的实验数据集也需要多元化以增强说服力。作者应在结论部分总结实验结果并提供更具体的未来展望。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

Contribution of the Submission

This paper is well written and easy to follow, contributing a scheme for proactive DeepFake defense approach. Before the streaming media files are released to the public network, the perturbations are added, thus making DeepFake invalid. This scheme performs well in the author’s paper experiment and can produce a good defense effect with nearly invisible disturbance.

Strength:

Identifies important issues in previous DeepFake detectors(especially in passive DeepFake detectors).
The proposed method adds perturbation to the HSV space, which has some certain innovations.

Weakness:

There are some problems, which must be solved before it is considered for publication. If the following problems are well-addressed, this reviewer believes that the essential contribution of this paper is important for De