README.md

CausIL

A repo for Causal Imitation Learning under Temporally Correlated Noise.

Running Experiments

To re-train an expert, run:

python experts/train_expert.py -e {lunarlander, halfcheetah, ant}

To train a learner, run:

jupyter notebook

and open up LunarLander.ipynb and PyBullet.ipynb. This package supports training via Behavioral Cloning, DoubIL, and ResiduIL.

Visualizing Results

Run:

jupyter notebook

and open up vis.ipynb.

Citing our Work

@article{swamy2021causal,
  author       = {Gokul Swamy and Sanjiban Choudhury and J. Andrew Bagnell and Zhiwei Steven Wu},
  title        = {Causal Imitation Learning under Temporally Correlated Noise},
  conference   = {Proceedings of the 39th International Conference on Machine Learning},
  url          = {https://arxiv.org/abs/2202.01312},
}