Safe Exploration in Finite Markov Decision Processes with Gaussian Processes.
Matteo TurchettaFelix BerkenkampAndreas KrausePublished in: NIPS (2016)
Keyphrases
- gaussian processes
- markov decision processes
- state and action spaces
- model based reinforcement learning
- interval estimation
- gaussian process
- optimal policy
- state space
- gaussian process regression
- reinforcement learning
- dynamic programming
- policy iteration
- decision theoretic planning
- transition matrices
- stationary policies
- average cost
- action space
- average reward
- finite number
- multi task
- markov decision process
- hyperparameters
- prior knowledge
- least squares
- machine learning
- decision problems
- upper bound
- model selection
- higher order