Task Agnostic Cost Prediction Module for Semantic Labeling in Active Learning.
Srikumar SastryNathan JacobsMariana BelgiuRaian Vargas MarettoPublished in: IGARSS (2023)
Keyphrases
- active learning
- active learning strategies
- cost sensitive
- prediction accuracy
- cost sensitive learning
- random sampling
- binary classification
- prediction model
- training set
- prediction algorithm
- class imbalance
- sample selection
- semantic information
- learning strategies
- prediction error
- labeled data
- semantic web
- domain specific
- selective sampling
- natural language
- learning algorithm
- training examples
- high cost
- machine learning
- expected cost
- high level
- image segmentation
- imbalanced data classification
- crowd sourced
- data sets
- labeling effort
- experimental design
- minimum cost
- semantic network
- semantic annotation
- semi supervised
- support vector machine
- knowledge base