Rethinking Supervised Learning and Reinforcement Learning in Task-Oriented Dialogue Systems.
Ziming LiJulia KiselevaMaarten de RijkePublished in: CoRR (2020)
Keyphrases
- dialogue system
- reinforcement learning
- supervised learning
- human computer
- dialogue management
- natural language
- natural language generation
- tutorial dialogue
- mixed initiative
- learning algorithm
- spoken dialogue systems
- unsupervised learning
- temporal difference
- state space
- learning tasks
- kernel based learning
- human users
- function approximation
- reinforcement learning algorithms
- language understanding
- learning problems
- semi supervised
- machine learning
- training data
- user model
- model free
- training set
- multi agent
- markov decision processes
- optimal policy
- labeled data
- transfer learning
- dynamic programming
- partially observable
- active learning