DRAMA at the PettingZoo: Dynamically Restricted Action Spaces for Multi-Agent Reinforcement Learning Frameworks.
Michael OesterleTim GramsChristian BarteltPublished in: HICSS (2024)
Keyphrases
- action space
- reinforcement learning
- learning frameworks
- multi agent
- single agent
- state space
- continuous state
- markov decision processes
- state and action spaces
- active learning
- multi agent systems
- skill learning
- kernel methods
- action selection
- cooperative
- learning algorithm
- semi supervised learning
- blended learning
- real valued
- unlabeled data
- optimal policy
- dynamic programming
- case study
- multiple agents
- unsupervised learning
- learning activities
- markov decision process
- online learning
- supervised learning
- image segmentation