Hindsight Reward Tweaking via Conditional Deep Reinforcement Learning.
Ning WeiJiahua LiangDi XieShiliang PuPublished in: CoRR (2021)
Keyphrases
- reinforcement learning
- action selection
- function approximation
- eligibility traces
- reinforcement learning algorithms
- optimal policy
- learning algorithm
- temporal difference
- markov decision processes
- state space
- multi agent
- average reward
- reward function
- learning process
- model free
- dynamic programming
- partially observable environments
- total reward
- machine learning
- random field model
- multi agent reinforcement learning
- supervised learning
- learning agent
- policy gradient
- multiscale
- belief nets
- probability distribution
- reward shaping
- conditional probabilities
- reinforcement learning methods
- policy evaluation
- transition model