README.md

Deep Reinforcement Learning Practice with Google Colab

Day 1: Value-Based RL

• 실습 교육 내용:
  • DQN
• 실습 내용:
  1. Prioritized Experience Replay (DQN+) [Schaul et al. ICLR 2016]
  2. Dueling network architectures for deep reinforcement learning (DQN++) [Wang et al. ICML 2016]

Day 2: Policy-Based RL

• 실습 교육 내용:
  1. REINFORCE
  2. Vanilla Policy Gradient (VPG) in the RLLAB library
• 실습 내용:
  • Deep Deterministic Policy Gradient (DDPG) [Lillicrap et al. ICLR 2016]

Day 3: Transfer Learning

• 실습 교육 내용:
  • Generative Adversarial Imitation Learning (GAIL) [Ho et al. NIPS 2016]
• 실습 내용:
  • Hyperparameter 조절해가면서 직접 실험.