LiteVSR: Efficient Visual Speech Recognition by Learning from Speech Representations of Unlabeled Data.
Hendrik LauxEmil MededovicAhmed HallawaLukas MartinArne PeineAnke SchmeinkPublished in: CoRR (2023)
Keyphrases
- unlabeled data
- supervised learning
- active learning
- learning algorithm
- labeled data
- semi supervised
- labeled and unlabeled data
- semi supervised learning
- co training
- domain adaptation
- positive examples
- learning process
- prior knowledge
- data points
- unsupervised learning
- training data
- neural network
- visual speech recognition
- text classification
- training samples
- speech recognition
- learning tasks
- pattern recognition
- feature extraction
- three dimensional
- labeled instances