Safe Reinforcement Learning as Wasserstein Variational Inference: Formal Methods for Interpretability.
Yanran WangDavid BoylePublished in: CoRR (2023)
Keyphrases
- formal methods
- variational inference
- reinforcement learning
- bayesian inference
- posterior distribution
- topic models
- probabilistic model
- gaussian process
- probabilistic graphical models
- mixture model
- variational methods
- latent dirichlet allocation
- model checking
- artificial intelligence
- formal specification
- closed form
- knowledge based systems
- exponential family
- state space
- parameter estimation
- graphical models
- factor graphs
- machine learning
- latent variables
- density estimation
- software engineering
- dynamic programming
- maximum a posteriori
- hyperparameters
- approximate inference
- learning algorithm
- natural language
- expert systems
- pairwise
- markov networks
- modeling language
- probability distribution
- prior information
- regression model