CrossAug: A Contrastive Data Augmentation Method for Debiasing Fact Verification Models.
Minwoo LeeSeungpil WonJuae KimHwanhee LeeCheon-Eum ParkKyomin JungPublished in: CIKM (2021)
Keyphrases
- synthetic data
- statistical methods
- data sets
- prior knowledge
- input data
- data collection
- test data
- experimental data
- detection method
- information loss
- historical data
- noisy data
- statistical model
- missing values
- clustering method
- dimensional vector
- segmentation method
- training samples
- statistical analysis
- machine learning methods
- prior information
- prediction model
- bayesian methods
- correlation analysis
- learned models
- missing data
- data analysis
- neural network
- face recognition
- training data
- preprocessing
- probabilistic model
- parameter estimation
- bayesian networks
- verification method
- large scale data sets
- contingency tables
- model selection
- model checking
- user input
- high quality
- parametric models
- probability distribution
- objective function
- incomplete data
- high dimensional data
- bayesian framework
- image data
- spectral clustering
- data sources
- data points