Feature selection with missing labels based on label compression and local feature correlation.
Lin JiangGuoxian YuMaozu GuoJun WangPublished in: Neurocomputing (2020)
Keyphrases
- feature selection
- multi label
- feature set
- text categorization
- class labels
- label space
- image labeling
- irrelevant features
- discriminative features
- feature subset
- redundant features
- multi label classification
- input features
- multi label learning
- multiple labels
- selecting features
- selection criterion
- label assignment
- labeling process
- mutual information
- missing values
- data compression
- feature space
- support vector
- multi instance learning
- multiple features
- compression scheme
- feature ranking
- machine learning
- compression ratio
- feature vectors
- training data
- cluster labels
- label noise
- text classification
- image compression
- image features
- multilabel classification
- supervised feature selection
- feature weighting
- feature selection algorithms
- compression algorithm
- model selection
- missing data
- label propagation
- labeled data
- semantic labels
- supervised learning
- image classification
- labeling effort
- feature extraction
- unlabeled data
- support vector machine
- classification accuracy
- lossless compression
- multi class
- pairwise