Let Offline RL Flow: Training Conservative Agents in the Latent Space of Normalizing Flows.
Dmitriy AkimovVladislav KurenkovAlexander NikulinDenis TarasovSergey KolesnikovPublished in: CoRR (2022)
Keyphrases
- latent space
- reinforcement learning
- latent variables
- low dimensional
- transfer learning
- high dimensional
- feature space
- generative model
- gaussian process
- learning algorithm
- state space
- training set
- gaussian processes
- dimensionality reduction
- search space
- distance metric
- parameter space
- pattern recognition
- manifold learning
- lower dimensional