Performance Improvement on Traditional Chinese Task-Oriented Dialogue Systems With Reinforcement Learning and Regularized Dropout Technique.
Jeng-Shin SheuSiang-Ru WuWen-Hung WuPublished in: IEEE Access (2023)
Keyphrases
- dialogue system
- reinforcement learning
- dialogue management
- human computer
- natural language
- tutorial dialogue
- mixed initiative
- natural language generation
- spoken dialogue systems
- state space
- function approximation
- machine learning
- language understanding
- reinforcement learning algorithms
- human users
- dialogue manager
- markov decision processes
- optimal policy
- knowledge engineering
- dynamic programming
- learning algorithm
- model free
- general purpose
- knowledge representation
- domain knowledge
- expert systems
- cooperative