#6 添加联邦学习模型参数加密功能

Closed
created 2 years ago by zhangyh02 · 3 comments
zhangyh02 added the
enhancement
label 2 years ago
qinsh was assigned by zhangyh02 2 years ago
qinsh commented 2 years ago
Collaborator
参考样例: AISynergy/examples/quickstart_pytorch 增加server端接口参数,进行配置加密模块以及压缩模块 syncore.server.run_server( ... mask_protocol='DHProtocol’ ) mask_protocol: 加密mask协议名称#TODO only Diffie–Hellman key excheange now 注意:加密mask正确性需保持正常连接(掉线功能#TODO)
zhangyh02 commented 2 years ago
Owner
DHProtocol加密协议每轮额外耗时多少?能否基于resnet50或盘古模型,展示测试结果
qinsh commented 2 years ago
Collaborator
额外耗时基本可以忽略,每轮迭代额外工作: 1 已经设置seed的随机数生成器生成mask 2 相加到参数上 测试resnet50,每轮迭代选择3个client mask duration(s): 0.03592681884765625 无损mask,对模型精度无影响
zhangyh02 closed this issue 2 years ago
Sign in to join this conversation.
No Milestone
No Assignees
2 Participants
Notifications
Due Date

No due date set.

Dependencies

This issue currently doesn't have any dependencies.

Loading…
There is no content yet.