The project is about using Reinforcement Learning to train an AI that can play Google Dino Game and compare the learning performance among different reward systems. A project poster is uploaded for a general overview for this project.
The Jupyter Notebook file only runs in a very specific working environment. Most likely, it will not run in your machine.
For simplicity, I will include a video file as a demonstration of my Jupyter Notebook. The video content will include following two things:
- Walkthrough of code
- Result demonstration
- YouTube link: https://youtu.be/Y6pUV4PqsYE
If you insisted to run through these codes by yourself，here is the guideline for you:
- All the package needed for this project is provided in the first part of the project as a pip install command.R
- Some newly installed package need to be rooted manually.
- In part 2.1, the observation window size and location need to be adjusted according to the screen size of your machine.
- You can test out your environment in part 2.2
- In part 3, while building the DQN model, the hyperparameter buffer_size need to be adjusted based the size of your machine's RAM. 420,000 is recommended for 16G RAM machine and 1,200,000 is recommended for 32G RAM machine.
Once you done with these, the Jupyter Notebook file should works for you. Enjoy.