Multilingual data selection for training stacked bottleneck features.
Ekapol ChuangsuwanichYu ZhangJames R. GlassPublished in: ICASSP (2016)
Keyphrases
- training data
- data sets
- training examples
- synthetic data
- data collection
- database
- prior knowledge
- training samples
- high quality
- data analysis
- raw data
- data points
- image features
- data quality
- original data
- structural information
- feature vectors
- classification accuracy
- data sources
- end users
- training dataset
- digital libraries
- feature space
- xml documents
- test data
- experimental data
- missing data
- information extraction
- labeled data
- knowledge discovery
- input data
- data mining techniques
- supervised learning
- small number
- image data