Generalized Expectation Criteria for Semi-Supervised Learning with Weakly Labeled Data
Abstract
In this paper, we present an overview of generalized expectation criteria (GE), a simple, robust, scalable method for semi-supervised training using weakly-labeled data. GE fits model parameters by favoring models that match certain expectation constraints, such as marginal label distributions, on the unlabeled data. This paper shows how to apply generalized expectation criteria to two classes of parametric models: maximum entropy models and conditional random fields. Experimental results demonstrate accuracy improvements over supervised training and a number of other state-of-the-art semi-supervised learning methods for these models.
Cite
Text
Mann and McCallum. "Generalized Expectation Criteria for Semi-Supervised Learning with Weakly Labeled Data." Journal of Machine Learning Research, 2010.Markdown
[Mann and McCallum. "Generalized Expectation Criteria for Semi-Supervised Learning with Weakly Labeled Data." Journal of Machine Learning Research, 2010.](https://mlanthology.org/jmlr/2010/mann2010jmlr-generalized/)BibTeX
@article{mann2010jmlr-generalized,
title = {{Generalized Expectation Criteria for Semi-Supervised Learning with Weakly Labeled Data}},
author = {Mann, Gideon S. and McCallum, Andrew},
journal = {Journal of Machine Learning Research},
year = {2010},
pages = {955-984},
volume = {11},
url = {https://mlanthology.org/jmlr/2010/mann2010jmlr-generalized/}
}