MILKDE: A new approach for multiple instance learning based on positive instance selection and kernel density estimation.
Alexandre W. C. FariaFrederico Gualberto F. CoelhoAlisson Marques da SilvaHonovan P. RochaGustavo Matheus de AlmeidaAndré P. LemosAntônio P. BragaPublished in: Eng. Appl. Artif. Intell. (2017)
Keyphrases
- instance selection
- multiple instance learning
- kernel density estimation
- density estimation
- supervised learning
- mean shift
- multi class
- probability density function
- object segmentation
- density function
- semi supervised
- image annotation
- semi supervised learning
- probability density
- text classification
- image processing
- image segmentation
- training set
- data reduction
- data compression
- computer vision
- class labels
- multi modal