Agent Modelling under Partial Observability for Deep Reinforcement Learning.
Georgios PapoudakisFilippos ChristianosStefano V. AlbrechtPublished in: NeurIPS (2021)
Keyphrases
- partial observability
- reinforcement learning
- partially observable
- learning agent
- agent programming
- belief state
- markov decision process
- state space
- symbolic model checking
- multi agent
- markov decision processes
- partial information
- fully observable
- reward function
- action selection
- learning algorithm
- dynamical systems
- reinforcement learning algorithms
- learning capabilities
- belief space
- planning problems
- function approximation
- temporal difference
- decision problems
- planning under partial observability
- multi agent systems
- infinite horizon
- multiagent systems
- single agent
- decision making
- partially observable markov decision processes
- solving problems
- optimal policy
- model free
- autonomous agents
- dynamic environments
- learning process