Risk-aware Q-learning for Markov decision processes.
Wenjie HuangWilliam Benjamin HaskellPublished in: CDC (2017)
Keyphrases
- markov decision processes
- risk sensitive
- policy iteration
- reinforcement learning
- reinforcement learning algorithms
- state space
- optimal policy
- markov games
- stochastic shortest path
- discounted reward
- reward function
- dynamic programming
- finite state
- transition matrices
- infinite horizon
- continuous state spaces
- discount factor
- reachability analysis
- factored mdps
- multi agent
- decision processes
- decision theoretic planning
- cooperative
- action space
- function approximation
- average reward
- finite horizon
- model free
- long run
- state action
- planning under uncertainty
- policy evaluation
- markov decision process
- action selection
- decision making
- learning algorithm
- model based reinforcement learning
- real time dynamic programming
- average cost
- partially observable
- action sets
- multistage
- dynamical systems
- actor critic
- markov decision problems