--
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-de...@googlegroups.com.
To post to this group, send email to cleverh...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/d7300526-51a0-4ff8-9d5c-6d5e433bd44c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
-NicolasThanks,Looking forward to your PR.Hi Vahid,This sounds like something that would be a good fit for the examples/ directory in the CleverHans repository.
On Tue, Jan 16, 2018 at 4:35 AM Vahid Behzadan <behz...@gmail.com> wrote:
Hi,--I have been working on an interface between Cleverhans and OpenAI Baselines for experimenting with adversarial example attacks against deep RL. The current version (available at https://github.com/behzadanksu/rl-attack ), supports FGSM, Iterative, and CW12 attacks on DQNs.First, I would like to ask whether the dev team would be interested in including this code to the main repository as an example of attacks on deep RL? This is a running project, and it would be great if I could continue its development in coordination and collaboration with other CleverHans developers.Second, my current implementation seem to run into memory issues in extremely extended experiments (e.g., >55M iterations for Breakthrough ). Do the FGSM module and its dependencies automatically clean up after generating the output, or do I have to do it manually? I welcome any advice on how to approach this problem.Many Thanks,Vahid
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-dev+unsubscribe@googlegroups.com.
To post to this group, send email to cleverhans-dev@googlegroups.com.
Thanks Nicolas, per our discussions in the PR thread, I am preparing an example-based version to avoid using submodules.Also, do you have any thoughts regarding my other question? Does the current implementation of FGSM clean up the memory after an example is generated? If not, would a customized clean up disturb its function at all?Cheers,
Vahid
On Tuesday, January 16, 2018 at 10:19:57 AM UTC-6, Nicolas Papernot wrote:
-NicolasThanks,Looking forward to your PR.Hi Vahid,This sounds like something that would be a good fit for the examples/ directory in the CleverHans repository.
On Tue, Jan 16, 2018 at 4:35 AM Vahid Behzadan <behz...@gmail.com> wrote:
Hi,--I have been working on an interface between Cleverhans and OpenAI Baselines for experimenting with adversarial example attacks against deep RL. The current version (available at https://github.com/behzadanksu/rl-attack ), supports FGSM, Iterative, and CW12 attacks on DQNs.First, I would like to ask whether the dev team would be interested in including this code to the main repository as an example of attacks on deep RL? This is a running project, and it would be great if I could continue its development in coordination and collaboration with other CleverHans developers.Second, my current implementation seem to run into memory issues in extremely extended experiments (e.g., >55M iterations for Breakthrough ). Do the FGSM module and its dependencies automatically clean up after generating the output, or do I have to do it manually? I welcome any advice on how to approach this problem.Many Thanks,Vahid
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-dev+unsubscribe@googlegroups.com.
To post to this group, send email to cleverh...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/d7300526-51a0-4ff8-9d5c-6d5e433bd44c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-dev+unsubscribe@googlegroups.com.
To post to this group, send email to cleverhans-dev@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/1bcc1592-bd2d-46c2-a349-d9ba2ac9432b%40googlegroups.com.
CleverHans doesn't do anything special in terms of memory management, but TensorFlow should garbage collect every buffer once it's done being used.
On Tue, Jan 23, 2018 at 3:19 PM, Vahid Behzadan <behz...@gmail.com> wrote:
Thanks Nicolas, per our discussions in the PR thread, I am preparing an example-based version to avoid using submodules.Also, do you have any thoughts regarding my other question? Does the current implementation of FGSM clean up the memory after an example is generated? If not, would a customized clean up disturb its function at all?Cheers,
Vahid
On Tuesday, January 16, 2018 at 10:19:57 AM UTC-6, Nicolas Papernot wrote:
-NicolasThanks,Looking forward to your PR.Hi Vahid,This sounds like something that would be a good fit for the examples/ directory in the CleverHans repository.
On Tue, Jan 16, 2018 at 4:35 AM Vahid Behzadan <behz...@gmail.com> wrote:
Hi,--I have been working on an interface between Cleverhans and OpenAI Baselines for experimenting with adversarial example attacks against deep RL. The current version (available at https://github.com/behzadanksu/rl-attack ), supports FGSM, Iterative, and CW12 attacks on DQNs.First, I would like to ask whether the dev team would be interested in including this code to the main repository as an example of attacks on deep RL? This is a running project, and it would be great if I could continue its development in coordination and collaboration with other CleverHans developers.Second, my current implementation seem to run into memory issues in extremely extended experiments (e.g., >55M iterations for Breakthrough ). Do the FGSM module and its dependencies automatically clean up after generating the output, or do I have to do it manually? I welcome any advice on how to approach this problem.Many Thanks,Vahid
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-dev+unsubscribe@googlegroups.com.
To post to this group, send email to cleverhans-dev@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/d7300526-51a0-4ff8-9d5c-6d5e433bd44c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Vahid
To post to this group, send email to cleverh...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/d7300526-51a0-4ff8-9d5c-6d5e433bd44c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-dev+unsubscribe@googlegroups.com.
To post to this group, send email to cleverh...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/1bcc1592-bd2d-46c2-a349-d9ba2ac9432b%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-dev+unsubscribe@googlegroups.com.
To post to this group, send email to cleverhans-dev@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/f01c9435-f7bb-473b-ab1c-f821f54d0fdf%40googlegroups.com.
Vahid
To post to this group, send email to cleverhans-dev@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/d7300526-51a0-4ff8-9d5c-6d5e433bd44c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "cleverhans dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cleverhans-dev+unsubscribe@googlegroups.com.
To post to this group, send email to cleverhans-dev@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cleverhans-dev/1bcc1592-bd2d-46c2-a349-d9ba2ac9432b%40googlegroups.com.