Ambiguity set construction for Model based Robust Reinforcement Learning. Many of the times, obtaining data for Reinforcement Learning problems is either difficult or expensive. This project proposes different approaches to construct ambiguity sets when the amount of data is not sufficient to build a precise model/transition kernel). This approach provides a robust solution with required performance guaranty when dealing with lack of data.
These instructions will get you a copy of the project up and running on your local machine for development and testing purposes.
Python 3
CRAAM (https://github.com/marekpetrik/CRAAM)
Gurobi (http://www.gurobi.com/)
- Follow the instructions in https://github.com/marekpetrik/CRAAM to in install CRAAM & the python interface of CRAAM.
- Install Gurobi from http://www.gurobi.com/
- Clone this repository. Run existing experiments from files Run_Experiments.py, Inventory_Optimization_Experiments.py.