Skip to content
Code accompanying the paper "Action Robust Reinforcement Learning and Applications in Continuous Control" https://arxiv.org/abs/1901.09184
Jupyter Notebook Python
Branch: master
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
models Hopper param noise pr_mdp part 3 Apr 14, 2019
Comparison_Plots.ipynb Commit files Apr 14, 2019
LICENSE Initial commit Apr 14, 2019
README.md Create README.md Apr 14, 2019
ddpg.py Commit files Apr 14, 2019
main.py Commit files Apr 14, 2019
network.py Commit files Apr 14, 2019
normalized_actions.py Commit files Apr 14, 2019
ounoise.py Commit files Apr 14, 2019
param_noise.py Commit files Apr 14, 2019
replay_memory.py
test.py Commit files Apr 14, 2019
utils.py Commit files Apr 14, 2019

README.md

Action-Robust-Reinforcement-Learning

Code accompanying the paper "Action Robust Reinforcement Learning and Applications in Continuous Control"

Requirements:

Howto train:

python3.6 main.py --updates_per_step 10 --env-name "Hopper-v2" --alpha 0.1 --method pr_mdp

Where method can take 3 values mdp pr_mdp or nr_mdp, where pr/nr are the probabilistic robust and noisy robust as defined in the paper.

All results are saved in the models folder.

Howto evaluate:

Once a model has been trained, run:

python3.6 test.py --eval_type model

where --eval_type model will evaluate for model (mass) uncertainty and --eval_type model_noise will create the 2d visualizations.

Howto visualize:

See Comparison_Plots.ipynb for an example of how to access and visualize your models.

You can’t perform that action at this time.