Project Index

Project Title: Playing Duck Game with Deep Reinforcement Learning

Project Team: Tiger Wang, Garrek Chan

Deliverables:

• Final report - Report.pdf (in this directory)
• Video demo - https://youtu.be/MJuzwlJ5XrQ
• Source code for modified Duck Game - https://github.com/thenewpotato/DuckGameRebuilt
• Source code for game environment wrapper and neural network implementation - https://github.com/thenewpotato/DuckQ (Also as DuckQ.ipynb in this directory)
• Trained reinforcement learning model (at 2600 episodes) artifact used for evaluation - models/duckq_net_1.chkpt in this repository

Instructions for Running:

• See Appendix A and Appendix B of Final Report PDF