Skip to content

Latest commit

 

History

History
20 lines (15 loc) · 920 Bytes

README.md

File metadata and controls

20 lines (15 loc) · 920 Bytes
Raw

Stable-oid

Implemented different RL algorithms to solve the infamous CartPole problem.

Algorithms:

  • Q-Learning:

    "Bucket-ised" the continous state space to construct a lookup table, a Q-table, which is used to perform updates as governed by the Bellman Optimality Equation. Check out q_learning_results.txt and the q_learning_plots folder for the write-up (on the complete training process) and plots, for consecutive runs, respectively.

    Insights from Ferdinand and Matthew.

  • Deep Q-Networks (DQN):

    (Coming soon)

To Do:

  • Tuning and update q_learning_results.txt
  • Implement DQN (with experience replay?)