Dysarthric Speech Recognition Using Pseudo-Labeling, Self-Supervised Feature Learning, and a Joint Multi-Task Learning Approach.
Ryoichi TakashimaYuya SawaRyo AiharaTetsuya TakiguchiYoshie ImaiPublished in: IEEE Access (2024)
Keyphrases
- multi task learning
- learning tasks
- multi task
- multiple tasks
- learning problems
- learning algorithm
- learning process
- supervised learning
- unsupervised learning
- multitask learning
- gaussian processes
- active learning
- theoretical analysis
- transfer learning
- inductive learning
- learning models
- expectation maximization
- metric learning
- image features
- feature vectors
- machine learning