Breaking the Beam Search Curse: A Study of (Re-)Scoring Methods and Stopping Criteria for Neural Machine Translation
Breaking the Beam Search Curse: A Study of (Re-)Scoring Methods and Stopping Criteria for Neural Machine Translation
复制标题
DOI：
10.18653/v1/d18-1342
复制
发表时间：
2018-08
期刊：
影响因子：
--
通讯作者：
Yilin Yang;Liang Huang;Mingbo Ma
中科院分区：
文献类型：
--
作者：
Yilin Yang;Liang Huang;Mingbo Ma
束搜索在神经机器翻译中应用广泛，与贪婪搜索相比，它通常能提升翻译质量。然而，人们普遍发现，束宽大于5时会损害翻译质量。我们解释了出现这种情况的原因，并提出了几种解决该问题的方法。此外，我们还讨论了这些方法的最优停止准则。结果表明，我们提出的无超参数方法比广泛使用的无超参数长度归一化启发式方法高出2.0个BLEU值，在汉译英任务的所有方法中取得了最佳效果。
Beam search is widely used in neural machine translation, and usually improves translation quality compared to greedy search. It has been widely observed that, however, beam sizes larger than 5 hurt translation quality. We explain why this happens, and propose several methods to address this problem. Furthermore, we discuss the optimal stopping criteria for these methods. Results show that our hyperparameter-free methods outperform the widely-used hyperparameter-free heuristic of length normalization by +2.0 BLEU, and achieve the best results among all methods on Chinese-to-English translation.