search

davide97l / rl-policies-attacks-defenses

Adversarial attacks on Deep Reinforcement Learning (RL)
MIT License
76 stars 12 forks source link

about paper link #23

Closed GongYanfu closed 1 year ago

GongYanfu commented 1 year ago

can you add the paper link of critical strategy attack? thanks 1664155800935

davide97l commented 1 year ago

Hello, critical strategy attack is an algorithm developed by me, that's why it can't be found online. However, you can read about it in my thesis work.

GongYanfu commented 1 year ago

thanks for your sharing

GongYanfu commented 1 year ago

and I have a serious question here: What do you think of the research direction of deep reinforcement learning-based attacks?

davide97l commented 1 year ago

That's a good question @GongYanfu, I think that given the recent achievements of the most advanced RL frameworks in outperforming previous algorithms in complex tasks (AlphaZero, MuZero, ...), similar frameworks could be used to craft more effective attacks. Those algorithms could both learn how to obfuscate observations with noise and to which states should attacked agents be led.

GongYanfu commented 1 year ago

but I find there are few papers about this research. can you tell me where and how did you find the related literature？