English

Sequence to Sequence Mixture Model for Diverse Machine Translation

Computation and Language 2018-10-18 v1

Abstract

Sequence to sequence (SEQ2SEQ) models often lack diversity in their generated translations. This can be attributed to the limitation of SEQ2SEQ models in capturing lexical and syntactic variations in a parallel corpus resulting from different styles, genres, topics, or ambiguity of the translation process. In this paper, we develop a novel sequence to sequence mixture (S2SMIX) model that improves both translation diversity and quality by adopting a committee of specialized translation models rather than a single translation model. Each mixture component selects its own training dataset via optimization of the marginal loglikelihood, which leads to a soft clustering of the parallel corpus. Experiments on four language pairs demonstrate the superiority of our mixture model compared to a SEQ2SEQ baseline with standard or diversity-boosted beam search. Our mixture model uses negligible additional parameters and incurs no extra computation cost during decoding.

Keywords

Cite

@article{arxiv.1810.07391,
  title  = {Sequence to Sequence Mixture Model for Diverse Machine Translation},
  author = {Xuanli He and Gholamreza Haffari and Mohammad Norouzi},
  journal= {arXiv preprint arXiv:1810.07391},
  year   = {2018}
}

Comments

11 pages, 5 figures, accepted to CoNLL2018

R2 v1 2026-06-23T04:42:45.041Z