rhklite/apprenticeship_inverse_RL

Apprenticeship Learning with Inverse Reinforcement Learning

Apprenticeship Learning with Inverse Reinforcement Learning

• Apprenticeship Learning with Inverse Reinforcement Learning

  • Deep Q Implementation
  • Tabular Q Implementation
  • Team Member

• This includes a double deep Q implementation (by Daniel. D) and a tabular Q method (by Richard H) of the paper P. Abbeel and A. Y. Ng, “Apprenticeship Learning via Inverse Reinforcement Learning. using CartPole model from openAI gym.

• Apprenticeship Learning via Inverse Reinforcement Learning.pdf is the presentation slides
• Apprenticeship_Inverse_Reinforcement_Learning.ipynb is the tabular Q implementation
• linearq.py is the deep Q implementation

• Running Colab: 1. File → playground mode, or Copy to Drive to open a copy 2. shift + enter to run 1 cell. Run all the cells
• Entire training takes around 10~15min

Deep Q Implementation

DQN Performance

Tabular Q Implementation

Tabular Q Performance

Team Member

• Richard Hu
• Daniel Dworakowski
• Omar Ismail