A Rate Distortion Approach for Semi-Supervised Conditional Random Fields

TitleA Rate Distortion Approach for Semi-Supervised Conditional Random Fields
Publication TypeConference Paper
Year of Publication2010
AuthorsG. Haffari, Y. Wang, Shaojun Wang, G. Mori
PublisherAdvances in Neural Information Processing Systems
Keywordssemi-supervised learning
Abstract

We propose a novel information theoretic approach for semi-supervised learning of conditional random fields that defines a training objective to combine the conditional likelihood on labeled data and the mutual information on unlabeled data. In contrast to previous minimum conditional entropy semi-supervised discriminative learning methods, our approach is grounded on a more solid foundation, the rate distortion theory in information theory. We analyze the tractability of the framework for structured prediction and present a convergent variational training algorithm to defy the combinatorial explosion of terms in the sum over label configurations. Our experimental results show the rate distortion approach outperforms standard l2 regularization, minimum conditional entropy regularization as well as maximum conditional entropy regularization on both multi-class classifcation and sequence labeling problems.

Full Text

Y. Wang, G. Haffari, S. Wang and G. Mori. (2010) A rate distortion approach for semi-supervised conditional random fields. in Yoshua Bengio And Dale Schuurmans (Eds.) Proceedings of the Twenty-Third Annual Conference on Neural Information Processing Systems, Advances in Neural Information Processing Systems, Volume 23, MIT Press.

Editor Yoshua Bengio And Dale Schuurmans (Eds.)