Towards Computationally Efficient Responsibility Attribution in Decentralized Partially Observable MDPs.
Stelios TriantafyllouGoran RadanovicPublished in: CoRR (2023)
Keyphrases
- partially observable
- computationally efficient
- dec pomdps
- markov decision processes
- infinite horizon
- markov decision problems
- state space
- reinforcement learning
- decision problems
- dynamical systems
- decentralized control
- multi agent
- computational complexity
- action models
- partial observability
- reward function
- dynamic programming
- partially observable markov decision processes
- belief state
- partially observable environments
- partial observations
- finite state
- optimal policy
- long run
- fully observable
- policy iteration
- planning domains
- probabilistic planning
- average cost
- function approximation
- continuous state
- reinforcement learning algorithms
- machine learning