Normalized Loss Functions for Deep Learning with Noisy Labels.
Xingjun MaHanxun HuangYisen WangSimone RomanoSarah M. ErfaniJames BaileyPublished in: CoRR (2020)
Keyphrases
- loss function
- deep learning
- pairwise
- unsupervised learning
- pairwise constraints
- loss minimization
- support vector
- machine learning
- risk minimization
- mental models
- learning to rank
- similarity measure
- training data
- weakly supervised
- class labels
- supervised learning
- multi label
- markov random field
- convex loss functions
- named entities
- higher order
- multiscale