Semi-Supervised Offline Reinforcement Learning with Action-Free Trajectories.
Qinqing ZhengMikael HenaffBrandon AmosAditya GroverPublished in: ICML (2023)
Keyphrases
- semi supervised
- reinforcement learning
- action selection
- partially observable domains
- action space
- semi supervised learning
- supervised learning
- reward shaping
- view invariant
- unlabeled data
- labeled data
- semi supervised classification
- unsupervised learning
- state action
- state space
- transition model
- function approximation
- machine learning
- semi supervised clustering
- multi view
- pairwise
- real time
- learning algorithm
- multi agent
- optimal policy
- motion patterns
- temporal difference
- transfer learning
- moving objects
- learning process
- active learning
- trajectory data
- model free
- learning agent
- dynamic programming
- sensory inputs
- co training
- action recognition