Policy Adaptation for Deep Reinforcement Learning-Based Dialogue Management.
Lu ChenCheng ChangZhi ChenBowen TanMilica GasicKai YuPublished in: ICASSP (2018)
Keyphrases
- dialogue management
- reinforcement learning
- partially observable markov decision process
- optimal policy
- partially observable
- learning agent
- dialogue system
- markov decision process
- state space
- action selection
- natural language generation
- markov decision processes
- partially observable markov decision processes
- reward function
- spoken dialogue systems
- action space
- function approximators
- decision theoretic
- language understanding
- markov decision problems
- function approximation
- learning algorithm
- belief state
- decision problems
- policy iteration
- infinite horizon
- dynamic programming
- multi agent
- learning capabilities
- long run
- reinforcement learning algorithms
- temporal difference
- transfer learning
- domain independent
- learning process
- virtual humans
- machine learning