Quality and Relevance Metrics for Selection of Multimodal Pretraining Data.
Roshan RaoSudha RaoElnaz NouriDebadeepta DeyAsli CelikyilmazBill DolanPublished in: CVPR Workshops (2020)
Keyphrases
- high quality
- data quality
- data sets
- training data
- data sources
- raw data
- data analysis
- image data
- database
- information retrieval
- complex data
- prior knowledge
- original data
- input data
- end users
- synthetic data
- data processing
- small number
- network structure
- human judgments
- data distribution
- missing data
- privacy preserving
- high dimensional data
- statistical analysis
- data mining techniques
- knowledge discovery
- data structure
- neural network