A Method for Stopping Active Learning Based on Stabilizing Predictions and the Need for User-Adjustable Stopping.
Michael BloodgoodK. Vijay-ShankerPublished in: CoRR (2014)
Keyphrases
- detection method
- significant improvement
- active learning
- cost function
- error rate
- clustering method
- experimental evaluation
- synthetic data
- preprocessing
- support vector machine
- pairwise
- user interface
- classification method
- high accuracy
- probabilistic model
- semi supervised learning
- computational cost
- user preferences
- high precision
- user input
- labeled data
- end users
- data sets
- prior knowledge
- evolutionary algorithm
- computational complexity
- learning environment
- feature selection
- genetic algorithm