State-Dependent Generalizations of Nonanticipatory Epsilon Entropy of Partially Observable Processes.
Charalambos D. CharalambousThemistoklis CharalambousPublished in: CDC (2022)
Keyphrases
- state dependent
- partially observable
- optimal policy
- markov decision processes
- decision problems
- infinite horizon
- state space
- reinforcement learning
- steady state
- dynamical systems
- partial observability
- queueing networks
- partial observations
- reward function
- stationary distribution
- finite state
- belief state
- continuous state
- sufficient conditions
- arrival rate
- dynamic programming
- machine learning
- long run
- single server
- markov chain
- computational complexity
- domain specific
- objective function
- service times
- markov decision process
- domain independent
- queue length
- orders of magnitude
- heuristic search