Distributed Structured Actor-Critic Reinforcement Learning for Universal Dialogue Management.
Zhi ChenLu ChenXiaoyuan LiuKai YuPublished in: CoRR (2020)
Keyphrases
- actor critic
- reinforcement learning
- dialogue management
- temporal difference
- policy gradient
- optimal control
- learning agent
- reinforcement learning algorithms
- dialogue system
- function approximation
- neuro fuzzy
- approximate dynamic programming
- gradient method
- multi agent
- spoken dialogue systems
- natural language generation
- policy iteration
- model free
- language understanding
- evaluation function
- learning process
- average reward
- state space
- machine learning
- supervised learning
- markov decision processes
- learning algorithm
- dynamic programming
- partially observable
- virtual humans
- rl algorithms
- dynamical systems
- partially observable markov decision process