OASIS: Conditional Distribution Shaping for Offline Safe Reinforcement Learning.
Yihang YaoZhepeng CenWenhao DingHaohong LinShiqi LiuTingnan ZhangWenhao YuDing ZhaoPublished in: CoRR (2024)
Keyphrases
- conditional distribution
- reinforcement learning
- reward shaping
- random variables
- gaussian process
- joint distribution
- probability distribution
- marginal distributions
- latent variable models
- state space
- posterior distribution
- dirichlet process
- learning process
- learning algorithm
- conditional probabilities
- hyperparameters
- random fields
- posterior probability
- cross validation
- model selection
- graphical models
- supervised learning
- k means
- multiscale