Exploration vs Exploitation with Partially Observable Gaussian Autoregressive Arms.
Julia KuhnMichel MandjesYoni NazarathyPublished in: VALUETOOLS (2014)
Keyphrases
- autoregressive
- partially observable
- decision problems
- state space
- dynamical systems
- non stationary
- reinforcement learning
- markov decision processes
- gaussian markov random field
- partial observability
- infinite horizon
- random fields
- markov decision problems
- partial observations
- belief state
- multiresolution
- partially observable markov decision processes
- maximum likelihood
- sar images
- optimal policy
- model selection
- sufficient conditions
- reward function
- machine learning