A set of simple problems solved using (D)RL

Task	Demo
CartPole-v1 using A2C
MountainCar-v0 using DQN
MountainCar-v0 using a custom DQN
MountainCar-v0 using a custom ConvolutionalDQN
LunarLander-v2 using a custom DQN
LunarLander-v2 using a custom ConvolutionalDQN

1. Nothing so far

1. Feature: Implement a custom DQN to solve the MountainCar_v0 environment
2. Feature: Implement a custom policy-based model to solve the CartPole_v1 environment
3. Feature: Implement a custom A2C model to solve the CartPole_v1 environment
4. Feature: Implement a vision-based DQN to solve the MountainCar_v0 enviroment
5. Feature: Implement a vision-based A2C to solve the CartPole_v1 enviroment
6. Feature: Solve Moon Lander using a custom agent
7. Feature: Add model architecture visualization to each project
8. Feature: Add train and test accuracy plots for each project
9. Feature: Add performance benchmark to all projects
10. Feature: Solve MountainCar-v0 without reward-shaping
11. Feature: Implement Prioritized Experience Replay (PER)
12. Feature: Solve some of the MuJoCo environments

1. Hamid Mohammadi: [email protected]