An end-to-end approach to combine attention feature extraction and Gaussian Process models for deep multiple instance learning in CT hemorrhage detection.
Jose Pérez-CanoYunan WuArne SchmidtMiguel López-PérezPablo Morales-ÁlvarezRafael MolinaAggelos K. KatsaggelosPublished in: Expert Syst. Appl. (2024)
Keyphrases
- end to end
- multiple instance learning
- feature extraction
- multiple instance
- multi class
- gaussian process models
- text localization and recognition
- gaussian processes
- image annotation
- gaussian process
- semi supervised
- object detection
- image classification
- supervised learning
- semi supervised learning
- feature vectors
- machine learning
- image processing
- dimensionality reduction
- active learning
- high resolution
- pairwise
- feature space
- feature selection