Learning Rewards to Optimize Global Performance Metrics in Deep Reinforcement Learning.
Junqi QianPaul WengChenmien TanPublished in: CoRR (2023)
Keyphrases
- reinforcement learning
- learning problems
- learning process
- learning algorithm
- learning tasks
- learning capabilities
- function approximation
- learning agents
- active learning
- online learning
- multi agent
- reinforcement learning methods
- continuous state
- transfer learning
- unsupervised learning
- dynamic programming
- complex domains
- learned knowledge
- evolutionary learning
- multi agent reinforcement learning
- actor critic
- deep architectures
- eligibility traces