This repository contains the code files that we used in our work to construct two environments and test three learning methods. The environments in this repository:
- Driving simulator
- Dynamic treatment regime
To recreate our simulations firstly clone this git and then run the scripts specified below, each one from its own path.
- Dynamic treatment regime/Linear/Ellipsoid/ellipsoid_medical.py
- Dynamic treatment regime/Linear/Blackbox_loss1/bb_medical.py
- Dynamic treatment regime/Linear/Blackbox_loss2/bbl2_medical.py
- Driving simulation/Linear/Ellipsoid/ellipsoid_driving.py
- Driving simulation/Linear/Blackbox_loss1/bb_driving.py
- Driving simulation/Linear/Blackbox_loss2/bbl2_driving.py
- Driving simulation/non_linear/Ellipsoid/ellipsoid_non_linear.py
- Driving simulation/non_linear/Blackbox_loss2/bb_non_linear.py
- use the jupyter notebooks in each environment.
in this work we use the processed data from point85AI git repository that can be found at:
https://github.com/point85AI/Policy-Iteration-AI-Clinician
- Policy-iteration-AI-Clinician/data/normalized_data.mat
- Dynamic treatment regime/data/normalized_data.mat