Safe Imitation Learning via Fast Bayesian Reward Inference from Preferences.
Daniel S. BrownRussell ColemanRavi SrinivasanScott NiekumPublished in: ICML (2020)
Keyphrases
- imitation learning
- reinforcement learning
- bayesian inference
- bayesian networks
- maximum margin
- robotic systems
- relational domains
- humanoid robot
- posterior distribution
- posterior probability
- probabilistic inference
- learning algorithm
- markov decision processes
- support vector
- statistical relational learning
- average reward
- exact inference
- decision theory
- markov networks
- concept learning
- dynamic programming
- belief networks
- relational data
- latent variables
- transfer learning
- maximum likelihood
- collaborative filtering