Risk-sensitive reinforcement learning: a martingale approach to reward uncertainty.
Nelson VadoriSumitra GaneshPrashant P. ReddyManuela VelosoPublished in: ICAIF (2020)
Keyphrases
- risk sensitive
- reinforcement learning
- model free
- optimal control
- markov decision processes
- markov decision problems
- control policies
- expected utility
- function approximation
- state space
- utility function
- reinforcement learning algorithms
- temporal difference
- markov decision chains
- reward function
- average reward
- optimal policy
- multi agent
- dynamic programming
- decision theory
- partially observable
- decision theoretic
- policy iteration
- control policy
- action space
- control strategies
- average cost
- finite state
- learning capabilities
- markov decision process
- machine learning
- supervised learning
- finite horizon
- learning agent
- infinite horizon
- belief functions
- planning problems
- risk averse
- decision making
- learning algorithm