Meta-Learning for Improving Rare Word Recognition in End-to-End ASR.
Florian LuxNgoc Thang VuPublished in: ICASSP (2021)
Keyphrases
- end to end
- meta learning
- word recognition
- automatic speech recognition
- learning tasks
- inductive learning
- noisy environments
- speech recognition
- cursive script
- character recognition
- model selection
- handwriting recognition
- machine learning algorithms
- hidden markov models
- word segmentation
- decision trees
- document analysis
- speech signal
- base classifiers
- machine learning
- data mining
- metamodel
- transfer learning
- data model
- training data
- noise reduction
- feature selection
- real world
- neural network