Debias NLU Datasets via Training-free Perturbations.
Qi GuoYuanhang TangYawen OuyangZhen WuXinyu DaiPublished in: EMNLP (Findings) (2023)
Keyphrases
- training dataset
- benchmark datasets
- neural network
- uci machine learning repository
- linear svm
- training samples
- test data
- data sets
- training phase
- training algorithm
- training process
- database
- raw data
- test set
- supervised learning
- probabilistic model
- data structure
- feedforward neural networks
- information systems
- computer software