Reward Shaping via Diffusion Process in Reinforcement Learning.
Peeyush KumarPublished in: CoRR (2023)
Keyphrases
- reward shaping
- diffusion process
- reinforcement learning
- reinforcement learning algorithms
- complex domains
- diffusion processes
- anisotropic diffusion
- state space
- function approximation
- markov decision problems
- homogeneous areas
- learning algorithm
- reward function
- machine learning
- queue length
- flow field
- optimal policy
- model free
- action selection
- markov decision processes
- continuous state
- transition model
- policy search
- temporal difference
- dynamical systems
- markov decision process
- edge detection
- agent learns
- supervised learning