LAVA: Latent Action Spaces via Variational Auto-encoding for Dialogue Policy Optimization.
Nurul LubisChristian GeishauserMichael HeckHsien-Chin LinMarco MoresiCarel van NiekerkMilica GasicPublished in: COLING (2020)
Keyphrases
- action space
- state space
- state and action spaces
- markov decision processes
- real valued
- reinforcement learning
- control policies
- reinforcement learning problems
- continuous state
- skill learning
- action selection
- markov decision process
- image segmentation
- reinforcement learning algorithms
- state action
- continuous state spaces
- policy search
- dynamic programming
- machine learning
- latent variables
- optimal policy
- search algorithm