Per-Step Reward: A New Perspective for Risk-Averse Reinforcement Learning.
Shangtong ZhangBo LiuShimon WhitesonPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- risk averse
- risk neutral
- decision makers
- function approximation
- utility function
- risk aversion
- model free
- state space
- reinforcement learning algorithms
- learning algorithm
- machine learning
- dynamic programming
- stochastic programming
- optimal policy
- expected utility
- random variables
- markov decision processes
- reward function
- control policy
- average reward