A Game-Theoretic Perspective on Risk-Sensitive Reinforcement Learning.
Mathieu GodboutMaxime HeuilletSharath Chandra RaparthyRupali BhatiAudrey DurandPublished in: SafeAI@AAAI (2022)
Keyphrases
- risk sensitive
- game theoretic perspective
- reinforcement learning
- model free
- optimal control
- markov decision processes
- markov decision problems
- control policies
- reinforcement learning algorithms
- state space
- utility function
- function approximation
- optimal policy
- temporal difference
- markov decision chains
- finite state
- policy iteration
- machine learning
- dynamic programming
- average cost
- action space
- learning algorithm
- control strategies
- markov decision process
- infinite horizon
- decision theoretic
- long run
- multi objective