HIQL: Offline Goal-Conditioned RL with Latent States as Actions.
Seohong ParkDibya GhoshBenjamin EysenbachSergey LevinePublished in: CoRR (2023)
Keyphrases
- reinforcement learning
- perceptual aliasing
- initial state
- goal state
- state space
- action selection
- state transitions
- state action
- action sequences
- initially unknown
- multiagent reinforcement learning
- optimal policy
- agent learns
- partial knowledge
- real time
- learning algorithm
- machine learning
- state information
- reward signal
- learned knowledge
- action space
- average reward
- reinforcement learning algorithms
- decision theoretic
- function approximation
- markov decision processes
- video sequences