cleverhans dev

Contact owners and managers

1–30 of 66

cleverhans is a library for benchmarking vulnerability to adversarial examples. You can find it there: https://github.com/openai/cleverhans

0 selected

Andres Antillon, Nicholas Carlini3

8/6/19

Experimental Public Version of Cleverhans

In the README there are steps for installing from the current HEAD: > If you'd instead like to

unread,

Experimental Public Version of Cleverhans

In the README there are steps for installing from the current HEAD: > If you'd instead like to

8/6/19

Cassidy Laidlaw, Nicolas Papernot4

7/11/19

SPSA in PyTorch

Can you raise an issue in the tracker and tag Jonathan? On Thu, Jul 11, 2019 at 12:05 PM Cassidy

unread,

SPSA in PyTorch

Can you raise an issue in the tracker and tag Jonathan? On Thu, Jul 11, 2019 at 12:05 PM Cassidy

7/11/19

Ian Goodfellow, Nicholas Carlini2

1/4/19

drop python 2.7 support

Yes. Python3 is is 10 years old now. The world should move on. Besides, Python2.7 will officially EOL

unread,

drop python 2.7 support

Yes. Python3 is is 10 years old now. The world should move on. Besides, Python2.7 will officially EOL

1/4/19

ephi...@yahoo.com, Nicholas Carlini3

12/26/18

implementing Carlini-Wagner attack

Yes. Thank you. Your reply is very helpful. And I apologize that I misunderstood and have been using

unread,

implementing Carlini-Wagner attack

Yes. Thank you. Your reply is very helpful. And I apologize that I misunderstood and have been using

12/26/18

ephi...@yahoo.com, Ian Goodfellow3

12/21/18

targeted or untargeted

Thank you for your rapid response and expert help. AT On Tuesday, December 18, 2018 at 3:13:57 PM UTC

unread,

targeted or untargeted

Thank you for your rapid response and expert help. AT On Tuesday, December 18, 2018 at 3:13:57 PM UTC

12/21/18

ephi...@yahoo.com, Ian Goodfellow2

12/13/18

factory = model.dataset_factory

cleverhans.dataset In the script where you make the trained model, you want to have a line like:

unread,

factory = model.dataset_factory

cleverhans.dataset In the script where you make the trained model, you want to have a line like:

12/13/18

ephi...@yahoo.com, Ian Goodfellow8

12/8/18

How do I evaluate tf.sign(grad)?

Thank you. I am looking forward to applying this update asap Monday. FYI, I received an email

unread,

How do I evaluate tf.sign(grad)?

Thank you. I am looking forward to applying this update asap Monday. FYI, I received an email

12/8/18

ephi...@yahoo.com, Ian Goodfellow3

11/27/18

Create a new model and train it to be robust to Fast Gradient Method attack

Thank you. On Tuesday, November 27, 2018 at 6:44:09 PM UTC-5, Ian Goodfellow wrote: This part is just

unread,

Create a new model and train it to be robust to Fast Gradient Method attack

Thank you. On Tuesday, November 27, 2018 at 6:44:09 PM UTC-5, Ian Goodfellow wrote: This part is just

11/27/18

ephi...@yahoo.com, Ian Goodfellow7

11/27/18

visualize the perturbation

In cleverhans_tutorials\evaluate_pickled_model.py I have added adv_x_sample = adv_x.eval(session=sess

unread,

visualize the perturbation

In cleverhans_tutorials\evaluate_pickled_model.py I have added adv_x_sample = adv_x.eval(session=sess

11/27/18

Nicolas Papernot

11/20/18

releasing CleverHans 3.0

I'm preparing the 3.0 release now that all open PRs have been merged. If anyone has objections

unread,

releasing CleverHans 3.0

I'm preparing the 3.0 release now that all open PRs have been merged. If anyone has objections

11/20/18

ephi...@yahoo.com, … Ian Goodfellow13

11/19/18

are eps values hard coded?

will do. thank you. On Monday, November 19, 2018 at 2:39:52 PM UTC-5, Ian Goodfellow wrote: I'd

unread,

are eps values hard coded?

will do. thank you. On Monday, November 19, 2018 at 2:39:52 PM UTC-5, Ian Goodfellow wrote: I'd

11/19/18

ephi...@yahoo.com, … Ian Goodfellow6

11/8/18

saving/loading CNN weights and losses to run different attack algorithms

Will do. Thank you for your expert help. Arnold On Thursday, November 8, 2018 at 4:03:25 PM UTC-5,

unread,

saving/loading CNN weights and losses to run different attack algorithms

Will do. Thank you for your expert help. Arnold On Thursday, November 8, 2018 at 4:03:25 PM UTC-5,

11/8/18

ephi...@yahoo.com, Ian Goodfellow3

11/1/18

No module named 'joblib'

Thank you for your rapid response. Problem is fixed. Sorry... I cleared my screen after installing

unread,

No module named 'joblib'

Thank you for your rapid response. Problem is fixed. Sorry... I cleared my screen after installing

11/1/18

Ian Goodfellow, Nicolas Papernot2

10/26/18

Support or users mailing list?

We've decided to have support requests made directly on stack overflow for now, see https://

unread,

Support or users mailing list?

We've decided to have support requests made directly on stack overflow for now, see https://

10/26/18

AkashGanesan, Ian Goodfellow2

10/25/18

Preprocessing as a step to attacks

The easiest way to do this is to just put it in Model.fprop. You can write Models that wrap other

unread,

Preprocessing as a step to attacks

The easiest way to do this is to just put it in Model.fprop. You can write Models that wrap other

10/25/18

ephi...@yahoo.com, Ian Goodfellow4

10/22/18

How do I implement alternate adversarial attacks in the tutorials

Terrific. Thank you for your help and rapid response On Monday, October 22, 2018 at 12:57:28 PM UTC-4

unread,

How do I implement alternate adversarial attacks in the tutorials

Terrific. Thank you for your help and rapid response On Monday, October 22, 2018 at 12:57:28 PM UTC-4

10/22/18

10/2/18

Paid Travis subscription

I set up a paid Travis subscription, or at least I tried to. Does anyone know how to tell if it

unread,

Paid Travis subscription

I set up a paid Travis subscription, or at least I tried to. Does anyone know how to tell if it

10/2/18

Ian Goodfellow, Nicolas Papernot4

9/28/18

Visualization and plotting module

https://github.com/tensorflow/cleverhans/pull/641 On Fri, Sep 28, 2018 at 11:22 AM, Nicolas Papernot

unread,

Visualization and plotting module

https://github.com/tensorflow/cleverhans/pull/641 On Fri, Sep 28, 2018 at 11:22 AM, Nicolas Papernot

9/28/18

Ian Goodfellow, Nicolas Papernot3

9/28/18

https://github.com/tensorflow/cleverhans/pull/640 On Fri, Sep 28, 2018 at 11:21 AM, Nicolas Papernot

unread,

https://github.com/tensorflow/cleverhans/pull/640 On Fri, Sep 28, 2018 at 11:21 AM, Nicolas Papernot

9/28/18

Yash Sharma, … Ian Goodfellow6

7/20/18

Ensemble Attacks

Sure! That's different than calculating the losses for each model and averaging, but still works.

unread,

Ensemble Attacks

Sure! That's different than calculating the losses for each model and averaging, but still works.

7/20/18

Ian Goodfellow, Nicolas Papernot2

7/10/18

I can't remember of any reasons we would have selected /tmp instead of the tutorials working

unread,

I can't remember of any reasons we would have selected /tmp instead of the tutorials working

7/10/18

Ian Goodfellow, … Yash Sharma15

7/9/18

Re: Google Scholar Problems

Awesome! Thanks Ian. On Mon, Jul 9, 2018 at 6:59 PM, Ian Goodfellow <goodfel...@gmail.com>

unread,

Re: Google Scholar Problems

Awesome! Thanks Ian. On Mon, Jul 9, 2018 at 6:59 PM, Ian Goodfellow <goodfel...@gmail.com>

7/9/18

Aidan Gomez, … Nicolas Papernot4

7/3/18

Transitioning away from PEP8

Transitioning to 2 spaces instead of 4 would be reasonable to be consistent with the overall

unread,

Transitioning away from PEP8

Transitioning to 2 spaces instead of 4 would be reasonable to be consistent with the overall

7/3/18

Aidan Gomez, Nicolas Papernot2

7/3/18

Parameter Naming Scheme

LGTM, thanks Aidan! On Tue, Jul 3, 2018 at 8:52 AM Aidan Gomez <ai...@cs.toronto.edu> wrote:

unread,

Parameter Naming Scheme

LGTM, thanks Aidan! On Tue, Jul 3, 2018 at 8:52 AM Aidan Gomez <ai...@cs.toronto.edu> wrote:

7/3/18

Ian Goodfellow, neuro...@gmail.com2

6/18/18

Fwd: Member requesting to join cleverhans dev

Thanks for the response Ian, I'll stay tuned and give as much as my capacity allows. Am Montag,

unread,

Fwd: Member requesting to join cleverhans dev

Thanks for the response Ian, I'll stay tuned and give as much as my capacity allows. Am Montag,

6/18/18

Yash Sharma, … Nicolas Papernot17

6/18/18

Cleverhans 2.1 Release - Finalization

This is now complete: https://github.com/tensorflow/cleverhans/releases/tag/v.2.1.0 On Fri, Jun 15,

unread,

Cleverhans 2.1 Release - Finalization

This is now complete: https://github.com/tensorflow/cleverhans/releases/tag/v.2.1.0 On Fri, Jun 15,

6/18/18

Yash Sharma, … Nicolas Papernot5

5/17/18

Madry Challenge C&W attack (name scopes)

This sounds like it would be a great addition to the examples folder. There is already a folder there

unread,

Madry Challenge C&W attack (name scopes)

This sounds like it would be a great addition to the examples folder. There is already a folder there

5/17/18

Arkadiusz Warzyński, Nicolas Papernot4

3/21/18

Limitations for features

I have read the article and I see one difference. My model contains both binary and numerical values

unread,

Limitations for features

I have read the article and I see one difference. My model contains both binary and numerical values

3/21/18

谢慈航, Ian Goodfellow2

1/25/18

implementation of L-BFGS based attack

Sounds good to me, thanks On Wed, Jan 24, 2018 at 11:46 PM, 谢慈航 <cihang...@gmail.com> wrote:

unread,

implementation of L-BFGS based attack

Sounds good to me, thanks On Wed, Jan 24, 2018 at 11:46 PM, 谢慈航 <cihang...@gmail.com> wrote:

1/25/18

Vahid Behzadan, … Ian Goodfellow7

1/23/18

Inclusion of framework for adversarial example attacks on deep RL - question about memory usage.

Makes perfect sense Ian, thanks. I'll check to see if this actually is the case and will take it

unread,

Inclusion of framework for adversarial example attacks on deep RL - question about memory usage.

Makes perfect sense Ian, thanks. I'll check to see if this actually is the case and will take it

1/23/18

Search

Clear search

Close search

Google apps

Main menu