MSA-MIL: A deep residual multiple instance learning model based on multi-scale annotation for classification and visualization of glomerular spikes.
Yilin ChenMing LiYongfei WuXueyu LiuFang HaoDaoxiang ZhouXiaoshuang ZhouChen WangPublished in: CoRR (2020)
Keyphrases
- multiple instance learning
- image categorization
- supervised learning
- image annotation
- multiple instance
- multiscale
- instance selection
- class labels
- multi class
- semi supervised
- object based image retrieval
- decision trees
- active learning
- classification accuracy
- image understanding
- semi supervised learning
- support vector machine
- support vector
- diverse density
- visual features
- image processing
- training set
- text classification
- training data
- multi modal
- feature selection
- data sets
- localized content based image retrieval
- training samples
- model selection
- image classification
- feature vectors
- image retrieval
- feature space
- feature extraction
- learning algorithm