Learning from Training Dynamics: Identifying Mislabeled Data Beyond Manually Designed Features.
Qingrui JiaXuhong LiLei YuJiang BianPenghao ZhaoShupeng LiHaoyi XiongDejing DouPublished in: CoRR (2022)
Keyphrases
- prior knowledge
- data sets
- data analysis
- learning algorithm
- raw data
- learning systems
- small number
- training and testing data
- database
- learned models
- experimental data
- training samples
- data collection
- image data
- knowledge discovery
- data sources
- neural network
- training data
- data points
- recurrent networks
- machine learning
- human experts
- training examples
- xml documents
- hidden markov models
- data structure
- missing data
- spatial data
- data processing
- online learning
- feature selection
- noisy data
- extracted features
- input data
- feature set
- labelled data
- supervised learning