Reinforcement learning has many interesting applications. The following two videos from OpenAI are demonstrations of RL applied to robotics and gaming.
In assignment 3, we implemented the basics of neural networks. In this assignment, we will implement reinforcement learning algorithms and train agents on various interesting environments.
We make this assignment simple and fun so that you can spend more time on your final project.
Install the gym library:
pip install gym
Check out the documentation and play around with different environments.
Choose your favorite environment from https://gym.openai.com/envs/.
You can finish simple actor critic in the provided boilerplate here, or you can instead choose to implement an advanced RL algorithm, which might be necessary if you choose a hard environment.
Note: some environments may require installing additional packages. See https://github.com/openai/gym/.
Please submit to http://learn.tsinghua.edu.cn/, including:
CartPole-v1 # of episode :2100, avg score : 320.5
).