No Prior Mask: Eliminate Redundant Action for Deep Reinforcement Learning.
Dianyu ZhongYiqin YangQianchuan ZhaoPublished in: AAAI (2024)
Keyphrases
- eliminate redundant
- reinforcement learning
- action selection
- partially observable domains
- action space
- function approximation
- prior knowledge
- state action
- learning algorithm
- reward shaping
- machine learning
- markov decision processes
- transition model
- optimal policy
- state space
- deep learning
- prior information
- robotic control
- reasoning about actions
- partially observable
- temporal difference
- model free
- reward function
- probabilistic model
- continuous state
- agent learns
- multi agent
- bayesian networks
- optimal control
- learning problems
- fitted q iteration