Read the paper Playing Atari with Deep Reinforcement Learning (the paper is also inside the 'Papers' folder in the course materials), and implement a model that can play atari games.
The goals of this project are the following:
- Read and understand the paper.
- Add a brief summary of the paper at the start of the notebook.
- Mention and implement the preprocessing needed; you can add your own steps if needed.
- Load an Atari environment from OpenAI Gym; start with Pong, and try with at least one more.
- Define the convolutional model needed for training.
- Apply deep q learning with your model.
- Use the model to play a game and show the result.
Rubric:
- A summary of the paper was included. The summary covered what the paper does, and why, as well as the preprocessing steps and the model they introduced.
- Read images from the environment, and performed the correct preprocessing steps.
- Defined an agent class with the needed functions.
- Defined the model within the agent class.
- Trained the model with the Pong environment. Save the weights after each episode.
- Test the model by making it play Pong.
- Train and test the agent with another Atari environment of your choosing.
Deadline: 04/04/19