Abstract
Obtaining tagged training data takes a long time and also is a costly task. Active learning aims machine learning algorithms achieve reasonable accuracies with less tagged training data. To this purpose, one of the methods for determining which samples to be tagged is making use of the decisions of classifier ensembles. Within this work, we implemented a committee-based active learning application and compared it with non-active methods.
Keywords
Computer science
Active learning (machine learning)
Machine learning
Classifier (UML)
Artificial intelligence
Training set
Task (project management)
Selection (genetic algorithm)
Labeled data
Engineering
Subject Areas
Machine Learning and Algorithms
·Artificial Intelligence ·Physical Sciences
Machine Learning and Data Classification
·Artificial Intelligence ·Physical Sciences
Algorithms and Data Compression
·Artificial Intelligence ·Physical Sciences
Citations by Year
2014: 2
Loading…
Loading…