Pysyft: Create FL Demo with DP-SGD Optimizer

Created on 19 Feb 2019 · 6Comments · Source: OpenMined/PySyft

This is a sub-issue of the #1824 Epic.

Similar to the previous demo in the 0.3.1 version of the stack, we want to use the DP-SGD optimizer to train a differentially private federated learning model. This should use our SensitivityTensor, MomentsAccountant, and DP mechanisms previously implemented into PySyft for the budget calculation.

Status

Source

iamtrask

👍1

Most helpful comment

@karlhigley @Kritikalcoder looks like this issue isn't ready to be worked on yet. see https://github.com/OpenMined/PySyft/issues/1824 and in particular issue https://github.com/OpenMined/PySyft/issues/1932

Prtfw on 22 Mar 2020

👍2

All 6 comments

are there any updates on this?

dionman on 27 Aug 2019

👍2

Is anyone working on this?

Kritikalcoder on 22 Mar 2020

@Prtfw is probably the most knowledgeable about where we are with demos. Thoughts on whether/how to tackle this @Prtfw?

karlhigley on 22 Mar 2020

Prtfw on 22 Mar 2020

👍2

@trask What's the status on those issues? Are they still accurate/relevant?

karlhigley on 22 Mar 2020

This issue has been marked stale because it has been open 30 days with no activity. Leave a comment or remove the stale label to unmark it. Otherwise, this will be closed in 7 days.