Instance-adaptive training with noise-robust losses against noisy labels.
Lifeng JinLinfeng SongKun XuDong YuPublished in: EMNLP (1) (2021)
Keyphrases
- label noise
- noisy environments
- active learning
- image noise
- training set
- noisy data
- training examples
- measurement noise
- noise free
- training data
- noise model
- noise reduction
- pairwise
- low snr
- training process
- salt pepper
- high noise
- random noise
- decision trees
- greater robustness
- geometric distortions
- missing data
- test set
- training samples
- online learning
- annotation effort
- noise level
- imaging artifacts
- noisy measurements
- data sets
- multi label
- supervised learning
- e learning