/RL-Visual-Input-Game

Reinforcement learning algorithms for a visual input game

Primary LanguageJupyter NotebookMIT LicenseMIT

RL-visual-input

Reinforcement learning algorithms for a visual input task

1 Deep Deterministic Policy Gradients (DDPG)

  1. DDPG with vector input: ddpg_vec.ipynb
  2. DDPG with visual input: ddpg_vis.ipynb

2 Deep Q-Network (DQN)

  1. DQN with vector input: dqn_vec.ipynb
  2. DQN with visual input: dqn_vis.ipynb
  3. DQN + LSTM with visual input: dqn_lstm_vis.ipynb

Results can be found in report.pdf