Commit graph

378 commits

Author SHA1 Message Date
A. Unique TensorFlower
c5c807807f Add assert that the training is private.
In Keras training in TF 2.0+, compute_gradients() is not called but apply_gradients() is called. W/o calling compute_gradients() dp gradient is not computed, and a normal gradient is used.

PiperOrigin-RevId: 307822742
2020-04-22 08:35:41 -07:00
A. Unique TensorFlower
f7b4a30120 Merge pull request #89 from woodyx218:GDPrivacy
PiperOrigin-RevId: 307695867
2020-04-21 15:48:54 -07:00
Steve Chien
264820cb3f Automated rollback of changelist 306513264
PiperOrigin-RevId: 307694393
2020-04-21 15:38:12 -07:00
A. Unique TensorFlower
9627cc0ed8 COPYBARA_INTEGRATE_REVIEW=https://github.com/tensorflow/privacy/pull/89 from woodyx218:GDPrivacy d06340e1cf
PiperOrigin-RevId: 306534693
2020-04-15 08:31:10 -07:00
Shuang Song
0c2747462f Internal change.
PiperOrigin-RevId: 302557220
2020-03-23 17:41:25 -07:00
A. Unique TensorFlower
2301931725 Fix issue with importing tensorflow.compat.v1.
PiperOrigin-RevId: 300175680
2020-03-10 14:16:46 -07:00
A. Unique TensorFlower
6541960e79 Automated refactoring to make code Python 3 compatible.
PiperOrigin-RevId: 299961805
2020-03-09 16:19:53 -07:00
A. Unique TensorFlower
e002cc9a23 Automated refactoring to make code Python 3 compatible.
PiperOrigin-RevId: 299951850
2020-03-09 15:35:52 -07:00
A. Unique TensorFlower
fa9ed85f37 Automated refactoring to make code Python 3 compatible.
PiperOrigin-RevId: 299935369
2020-03-09 14:26:58 -07:00
A. Unique TensorFlower
f3672adf80 Explicitly disable Tensorflow v2 behaviors for all TF1.x binaries and tests
PiperOrigin-RevId: 299397784
2020-03-06 11:29:06 -08:00
Yanhua Sun
b0df24ef25 Explicitly replace "import tensorflow" with "tensorflow.compat.v1" for TF2.x migration
PiperOrigin-RevId: 297199727
2020-02-25 14:11:47 -08:00
woodyx218
b9b2e8670f
Move doc str below functions 2020-02-21 09:30:30 -05:00
Galen Andrew
945075a136 Fix num_records in quantile_adaptive_clip_sum_query_test.
PiperOrigin-RevId: 292995170
2020-02-07 10:19:08 -08:00
Galen Andrew
9bb3c1e6d8 A few new features for QuantileAdaptiveClipSumQuery.
1. Remove redundant global_state.l2_norm_clip from QuantileAdaptiveClipSumQuery.
2. Simplify accumulation code by deriving from SumAggregationDPQuery.
3. Add geometric update option to QuantileAdaptiveClipAverageQuery.

PiperOrigin-RevId: 292442733
2020-02-07 10:18:57 -08:00
Yanhua Sun
856eda3aa1 Explicitly replace "import tensorflow" with "tensorflow.compat.v1"
PiperOrigin-RevId: 290355582
2020-02-07 10:18:46 -08:00
woodyx218
7b72e8a11b
Add files via upload 2020-01-22 10:29:25 +08:00
woodyx218
2ef5c6e332
Add files via upload 2020-01-19 20:29:37 +08:00
woodyx218
0b01471497
Delete GDprivacy_accountants.py 2020-01-19 20:28:05 +08:00
Keith Rush
a8a2d91795 Move TF privacy off contrib.
PiperOrigin-RevId: 289953826
2020-01-15 15:47:21 -08:00
Galen Andrew
c80a862ae2 Remove tf.contrib.framework.
PiperOrigin-RevId: 289487098
2020-01-13 11:48:35 -08:00
woodyx218
cbce4540d3
Initial commit 2020-01-02 16:36:31 +08:00
Galen Andrew
8d98c3433b Add geometric updating for quantile-based adaptive clipping.
PiperOrigin-RevId: 285799976
2019-12-16 10:29:50 -08:00
Steve Chien
d69879d360 Changes to make Tensorflow Privacy compatible with TF 2.0.
PiperOrigin-RevId: 277561553
2019-10-30 12:33:02 -07:00
Steve Chien
2007aac912 Allow compute_dp_sgd_privacy to be called as library function.
PiperOrigin-RevId: 275966906
2019-10-21 18:10:05 -07:00
Steve Chien
1ce8cd4032 Restructure TF Privacy to be more in line with other repos in the TF ecosystem.
PiperOrigin-RevId: 274674077
2019-10-14 15:29:52 -07:00
Steve Chien
6c03ce49fd Add tensorflow_privacy/ to list of git_files.
PiperOrigin-RevId: 274022715
2019-10-10 13:09:04 -07:00
Steve Chien
b125e3a686 Add new top-level directory to GitHub repo, and add __init__.py file at top level. This makes the structure more consistent with other repos in the Google Tensorflow ecosystem.
PiperOrigin-RevId: 273803458
2019-10-09 13:15:10 -07:00
Steve Chien
313edfc80c Update setup.py to new release.
PiperOrigin-RevId: 273569076
2019-10-09 13:14:59 -07:00