TBC-MI : Suppressing noise labels by maximizing cleaning samples for robust image classification.
Yanhong LiZhiqing GuoLiejun WangLianghui XuPublished in: Inf. Process. Manag. (2024)
Keyphrases
- image classification
- image noise
- multi label
- salt pepper
- mutual information
- label noise
- noisy environments
- signal to noise ratio
- random noise
- data sets
- noise level
- impulse noise
- missing data
- image details
- robust statistical
- imaging artifacts
- image processing
- training samples
- image representation
- training set
- noise reduction
- noisy data
- median filter
- noise removal
- visual words
- bag of words
- visual features
- labeled samples
- image features
- pairwise
- training data