PeterLiu-all / peterliu-all.github.io

个人技术博客 https://peterliuzhi.top
https://peterliuzhi.top/
1 stars 0 forks source link

posts/%E8%AE%BA%E6%96%87%E7%AC%94%E8%AE%B0/%E9%9A%90%E7%A7%81%E4%BF%9D%E6%8A%A4/%E8%AE%BA%E6%96%87%E7%AC%94%E8%AE%B0fedml-he_an-efficient-homomorphic-encryption-based-privacy-preserving-federated-learning-system/ #3

Open utterances-bot opened 1 week ago

utterances-bot commented 1 week ago

【论文笔记】FEDML-HE_AN EFFICIENT HOMOMORPHIC-ENCRYPTION-BASED PRIVACY-PRESERVING FEDERATED LEARNING SYSTEM - P3troL1er 的个人博客

【论文笔记】FEDML-HE_AN EFFICIENT HOMOMORPHIC-ENCRYPTION-BASED PRIVACY-PRESERVING FEDERATED LEARNING SYSTEM-P3troL1er的个人技术博客

https://peterliuzhi.top/posts/%E8%AE%BA%E6%96%87%E7%AC%94%E8%AE%B0/%E9%9A%90%E7%A7%81%E4%BF%9D%E6%8A%A4/%E8%AE%BA%E6%96%87%E7%AC%94%E8%AE%B0fedml-he_an-efficient-homomorphic-encryption-based-privacy-preserving-federated-learning-system/

Red9th commented 1 week ago

您好,想问下您对于加密掩码图这里的计算是如何理解的。对客户端上传的加密敏感度矩阵进行同态聚合,得到的应该还是密文形式的聚合矩阵,那是如何按比率 p 从中选出最敏感的参数的? 是否应该先将聚合得到的加密敏感度矩阵发送给客户端,然后客户端对其解密之后,再按照比率 p 进行选择部分最敏感的参数,从而得到掩码图? 期待您的回复!