A meta-active learning approach exploiting instance importance.
Sergio FlescaDomenico MandaglioFrancesco ScalaAndrea TagarelliPublished in: Expert Syst. Appl. (2024)
Keyphrases
- active learning
- version space
- unlabeled instances
- machine learning
- random sampling
- selective sampling
- learning strategies
- relative importance
- learning algorithm
- supervised learning
- training examples
- semi supervised
- experimental design
- database
- learning process
- real world
- high level abstraction
- active learning strategies
- annotation effort
- cost sensitive
- pool based active learning
- imbalanced data classification
- music emotion classification
- crowd sourced
- meta level
- semi supervised learning
- domain knowledge
- website
- data sets