The First-Occupancy Representation

Implementation for our paper A First-Occupancy Representation for Reinforcement Learning.

dependencies: numpy, scipy, matplotlib, seaborn

files:

• agents.py: basic agent classes for GPI, value iteration, the FR, and the SR
• envs.py: FourRoom and Escape environment classes
• frp.py: FR planning (FRP) agent class
• utils.py: basic helper functions
• runners.py: functions for running experiments
• four_rooms.ipynb: result notebook

If you find this code useful, please cite using:

@misc{moskovitz2021firstoccupancy,
      title={A First-Occupancy Representation for Reinforcement Learning}, 
      author={Ted Moskovitz and Spencer R. Wilson and Maneesh Sahani},
      year={2021},
      eprint={2109.13863},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}