International Workshop on Spoken Language Translation (IWSLT) 2009

Tokyo, Japan
December 1-2, 2009

Structural Support Vector Machines for Log-Linear Approach in Statistical Machine Translation

Katsuhiko Hayashi (1), Taro Watanabe (2), Hajime Tsukada (2), Hideki Isozaki (2)

(1) Department of Information Science and Technology, University of Doshisha, Japan
(2) NTT Comunication Science Labolatories, Japan

Minimum error rate training (MERT) is a widely used learning method for statistical machine translation. In this paper, we present a SVM-based training method to enhance generalization ability. We extend MERT optimization by maximizing the margin between the reference and incorrect translations under the L2-norm prior to avoid overfitting problem. Translation accuracy obtained by our proposed methods is more stable in various conditions than that obtained by MERT. Our experimental results on the French- English WMT08 shared task show that degrade of our proposed methods is smaller than that of MERT in case of small training data or out-of-domain test data.

Full Paper     Presentation (pdf)

Bibliographic reference.  Hayashi, Katsuhiko / Watanabe, Taro / Tsukada, Hajime / Isozaki, Hideki (2009): "Structural support vector machines for log-linear approach in statistical machine translation", In IWSLT-2009, 144-151.