Contrastive Learning as Goal-Conditioned Reinforcement Learning.
Benjamin EysenbachTianjun ZhangRuslan SalakhutdinovSergey LevinePublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- learning process
- learning algorithm
- learning problems
- supervised learning
- data sets
- prior knowledge
- learning capabilities
- active exploration
- unsupervised learning
- learning systems
- function approximation
- model free
- learning scheme
- autonomous learning
- continuous state
- rl algorithms
- reinforcement learning methods
- temporal difference learning
- background knowledge
- dynamic programming
- active learning