Semi-supervised reward learning for offline reinforcement learning.
Ksenia KonyushkovaKonrad ZolnaYusuf AytarAlexander NovikovScott E. ReedSerkan CabiNando de FreitasPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- learning process
- supervised learning
- learning algorithm
- semi supervised
- learning agent
- active learning
- eligibility traces
- real time
- learning problems
- prior knowledge
- machine learning
- state space
- online learning
- function approximation
- temporal difference learning
- dynamic programming
- unsupervised learning
- learning tasks
- solving problems
- e learning
- reinforcement learning methods