Democratic co-learning
Authors:
- 22 February 2005
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 594-602
Abstract
For many machine learning applications it is important to develop algorithms that use both labeled and unlabeled data. We present democratic colearning in which multiple algorithms instead of multiple views enable learners to label data for each other. Our technique leverages off the fact that different learning algorithms have different inductive biases and that better predictions can be made by the voted majority. We also present democratic priority sampling, a new example selection method for active learning.Keywords
This publication has 14 references indexed in Scilit:
- Analyzing the effectiveness and applicability of co-training2000
- Text Classification from Labeled and Unlabeled Documents using EMMachine Learning, 2000
- Combining labeled and unlabeled data with co-training1998
- On the Accuracy of Meta-learning for Scalable Data MiningJournal of Intelligent Information Systems, 1997
- Committee-Based Sampling For Training Probabilistic Classifiers1995
- A Conservation Law for Generalization Performance1994
- A Sequential Algorithm for Training Text Classifiers1994
- Query by committee1992
- Induction of decision treesMachine Learning, 1986
- Nearest neighbor pattern classificationIEEE Transactions on Information Theory, 1967