Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add heu processor library files (for secure boost scheme phase 2) #10571

Open
wants to merge 2 commits into
base: vertical-federated-learning
Choose a base branch
from

Conversation

shaojian-ant
Copy link

This PR replaces [#10291 ], which changes the target branch from dmlc:master to dmlc:vertical-federated-learning

@shaojian-ant shaojian-ant changed the base branch from master to vertical-federated-learning July 11, 2024 12:58
@shaojian-ant shaojian-ant changed the title Vertical federated learning Add heu processor library files (for secure boost scheme phase 2) Jul 11, 2024
@shaojian-ant shaojian-ant marked this pull request as ready for review July 18, 2024 07:25
@shaojian-ant
Copy link
Author

Hi, @ZiyueXu77, @trivialfis,

I see that the Processor interface has been merged into the vertical-federated-learning branch. Given this progress, is it now ready for us to proceed with our previous proposal, which suggested integrating SecretFlow/HEU (https://github.com/secretflow/heu) as an alternative homomorphic encryption (HE) solution for Secure XGBoost?

To integrate our HEU processor implementation into XGBoost, should we provide the following three files?

  1. A dynamic library file (libproc_heu.so), placed in a specified directory in xgboost.
  2. A test file (test_*.cc), placed under the tests/cpp/processing directory.
  3. A README.md file, detailing information about our so file, including instructions on generating the public and secret keys.

Thanks for your guidance.

@trivialfis
Copy link
Member

Apologies for the lack of communication here. I ditched that branch as it's way too messy at this point. This is switched to https://github.com/dmlc/xgboost/tree/federated-secure . With the new branch, I believe there's no integration needed in XGBoost, rather, implementations will define an independent loadable library for XGBoost, and then provide the name of the library to XGBoost before training. I will try to provide a simple example as soon as possible.

The code is still in flux, as loading an external library in XGBoost has really not been expected since the beginning and we are still figuring out how to do it without messing up the rest of the world (hence not in the main branch). See the many limitations in #10410 (comment) .

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants