626_privacy/tensorflow_privacy
2
0
Fork 1
Code Issues Pull requests 1 Projects Releases Packages Wiki Activity Actions

trailing spaces

Browse source
This commit is contained in:
npapernot 2019-07-30 22:00:16 +00:00
parent cd0ff6d299
commit 223f2cc640
1 changed files with 5 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
privacy/bolton/README.md
View file

@ -1,15 +1,15 @@
# BoltOn Subpackage # BoltOn Subpackage
This package contains source code for the BoltOn method, a particular This package contains source code for the BoltOn method, a particular
differential-privacy (DP) technique that uses output perturbations and leverages differential-privacy (DP) technique that uses output perturbations and
additional assumptions to provide a new way of approaching the privacy leverages additional assumptions to provide a new way of approaching the
guarantees. privacy guarantees.
## BoltOn Description ## BoltOn Description
This method uses 4 key steps to achieve privacy guarantees: This method uses 4 key steps to achieve privacy guarantees:
1. Adds noise to weights after training (output perturbation). 1. Adds noise to weights after training (output perturbation).
2. Projects weights to R, the radius of the hypothesis space, 2. Projects weights to R, the radius of the hypothesis space,
after each batch. This value is configurable by the user. after each batch. This value is configurable by the user.
3. Limits learning rate 3. Limits learning rate
4. Use a strongly convex loss function (see compile) 4. Use a strongly convex loss function (see compile)