Skip to content
master
Switch branches/tags
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
adv
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Adversarial Attacks on Deep Reinforcement Learning

Installation

  1. Run git clone --recurse-submodules https://github.com/gkswamy98/adversarial_rl.git to clone with submodules and cd adversarial_rl.
  2. Run pip install -e . to install the main dependencies.
  3. Run mv baselines ../baselines, cd ../baselines, then pip install -e . --user to install our baselines fork.

Running Experiments

  • To train models, run bash scripts/train.sh.
    • To parallelize training runs, modify the training script to echo commands instead of executing them and run bash ../scripts/train.sh | xargs -PN -ICMD /bin/bash -exc CMD, replacing N with the number of cores available.
  • To attack trained models, run bash scripts/attack.sh.
    • Use the same trick as above to parallelize.

Results

PPO, TRPO, A2C and DQN models trained on CartPole and Acrobot are in the models/ folder. Results of attacks are available in accompanying writeup.

Authors

About

CS 294-131 Project, reimplementing https://arxiv.org/pdf/1702.02284.pdf.

Resources

Releases

No releases published

Packages

No packages published