Combining Reward Shaping and Curriculum Learning for Training Agents with High Dimensional Continuous Action Spaces.
Sooyoung JangMikyong HanPublished in: ICTC (2018)
Keyphrases
- action space
- multi agent
- learning process
- reinforcement learning
- cooperative
- multi agent systems
- learning algorithm
- skill learning
- learning agent
- single agent
- continuous action
- complex domains
- learning capabilities
- action selection
- hidden variables
- learning tasks
- domain independent
- dynamic programming
- prior knowledge