Memory Gym: Partially Observable Challenges to Memory-Based Agents in Endless Episodes.
Marco PleinesMatthias PallaschFrank ZimmerMike PreussPublished in: CoRR (2023)
Keyphrases
- partially observable
- partial observability
- multi agent systems
- markov decision processes
- multiple agents
- decision problems
- state space
- multi agent
- reinforcement learning
- reward function
- dynamical systems
- infinite horizon
- multiagent systems
- single agent
- partial observations
- belief state
- partially observable environments
- markov decision problems
- planning domains
- initially unknown
- action models
- partially observable domains
- planning problems
- decision making
- orders of magnitude
- machine learning
- reinforcement learning algorithms
- coalition formation
- linear programming
- special case