Top-Rank Enhanced Listwise Optimization for Statistical Machine Translation
Huadong Chen, Shujian Huang, David Chiang, Xinyu Dai, Jiajun Chen
Abstract
Pairwise ranking methods are the most widely used discriminative training approaches for structure prediction problems in natural language processing (NLP). Decomposing the problem of ranking hypotheses into pairwise comparisons enables simple and efficient solutions. However, neglecting the global ordering of the hypothesis list may hinder learning. We propose a listwise learning framework for structure prediction problems such as machine translation. Our framework directly models the entire translation list’s ordering to learn parameters which may better fit the given listwise samples. Furthermore, we propose top-rank enhanced loss functions, which are more sensitive to ranking errors at higher positions. Experiments on a large-scale Chinese-English translation task show that both our listwise learning framework and top-rank enhanced listwise losses lead to significant improvements in translation quality.- Anthology ID:
- K17-1011
- Volume:
- Proceedings of the 21st Conference on Computational Natural Language Learning (CoNLL 2017)
- Month:
- August
- Year:
- 2017
- Address:
- Vancouver, Canada
- Editors:
- Roger Levy, Lucia Specia
- Venue:
- CoNLL
- SIG:
- SIGNLL
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 90–99
- Language:
- URL:
- https://aclanthology.org/K17-1011
- DOI:
- 10.18653/v1/K17-1011
- Cite (ACL):
- Huadong Chen, Shujian Huang, David Chiang, Xinyu Dai, and Jiajun Chen. 2017. Top-Rank Enhanced Listwise Optimization for Statistical Machine Translation. In Proceedings of the 21st Conference on Computational Natural Language Learning (CoNLL 2017), pages 90–99, Vancouver, Canada. Association for Computational Linguistics.
- Cite (Informal):
- Top-Rank Enhanced Listwise Optimization for Statistical Machine Translation (Chen et al., CoNLL 2017)
- PDF:
- https://preview.aclanthology.org/proper-vol2-ingestion/K17-1011.pdf