Hybrid Reinforcement/Supervised Learning of Dialogue Policies from Fixed Data Sets.
James HendersonOliver LemonKallirroi GeorgilaPublished in: Comput. Linguistics (2008)
Keyphrases
- data sets
- supervised learning
- reinforcement learning
- training data
- training set
- optimal policy
- real world data sets
- natural language
- fitted q iteration
- markov decision process
- unsupervised learning
- active learning
- supervised machine learning
- man machine
- semi supervised
- learning algorithm
- machine learning
- multiple instance learning
- supervised classification
- learning tasks
- database
- learning problems
- benchmark data sets
- statistical learning
- hybrid learning
- neural network
- dialogue system
- labeled data
- tutorial dialogue
- generalization error
- human computer
- synthetic data
- training samples
- real world