Super-Prompting: Utilizing Model-Independent Contextual Data to Reduce Data Annotation Required in Visual Commonsense Tasks.
Navid RezaeiMarek Z. ReformatPublished in: CoRR (2022)
Keyphrases
- data sets
- experimental data
- data points
- raw data
- input data
- training data
- data analysis
- data sources
- prior knowledge
- probability distribution
- computer systems
- image data
- database
- data processing
- data distribution
- simulation data
- original data
- privacy preserving
- predictive model
- data collection
- high quality
- data mining techniques
- knowledge discovery
- statistical analysis
- synthetic data
- computational model
- end users
- probabilistic model
- artificial neural networks
- visual information
- high level
- prior information
- empirical data
- metadata
- learning algorithm