Source Domain Data Selection for Improved Transfer Learning Targeting Dysarthric Speech Recognition.
Feifei XiongJon BarkerZhengjun YueHeidi ChristensenPublished in: ICASSP (2020)
Keyphrases
- transfer learning
- speech recognition
- source domain
- target domain
- labeled data
- data sets
- domain adaptation
- hidden markov models
- machine learning
- unlabeled data
- data analysis
- databases
- cross domain
- knowledge transfer
- training data
- data distribution
- knowledge discovery
- reinforcement learning
- semi supervised learning
- language model
- learning tasks
- data points
- label information
- pattern recognition
- test data
- semi supervised
- data sources