Empirical Analysis of Beam Search Curse and Search Errors with Model Errors in Neural Machine Translation

Jianfei He, Shichao Sun, Xiaohua Jia, Wenjie Li

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

Abstract

Beam search is the most popular decoding method for Neural Machine Translation (NMT) and is still a strong baseline compared with the newly proposed sampling-based methods. To better understand the beam search, we investigate its two well-recognized issues, beam search curse and search error, not only on the test data as a whole but also at the sentence level. We find that only less than 30% of sentences in the WMT17 En–De and De–En test set experience these issues. Meanwhile, there is a related phenomenon. For the majority of sentences, their gold references get lower probabilities than the predictions from the beam search. We also test with different levels of model errors including a special test using training samples and models without regularization. In this test, the model has an accuracy of 95% in predicting the tokens on the training data. We find that these phenomena still exist even for such a model with very high accuracy. These findings show that it is not promising to improve the beam search by seeking higher probabilities and further reducing the search errors in decoding. The relationship between the quality and the probability at the sentence level in our results provides useful information to find new ways to improve NMT.

Original languageEnglish
Title of host publicationProceedings of the 24th Annual Conference of the European Association for Machine Translation, EAMT 2023
EditorsMary Nurminen, Mary Nurminen, Judith Brenner, Maarit Koponen, Sirkku Latomaa, Mikhail Mikhailov, Frederike Schierl, Tharindu Ranasinghe, Eva Vanmassenhove, Sergi Alvarez Vidal, Nora Aranberri, Mara Nunziatini, Carla Parra Escartin, Mikel Forcada, Maja Popovic, Carolina Scarton, Helena Moniz
PublisherEuropean Association for Machine Translation
Pages91-101
Number of pages11
ISBN (Electronic)9789520329471
Publication statusPublished - 2023
Event24th Annual Conference of the European Association for Machine Translation, EAMT 2023 - Tampere, Finland
Duration: 12 Jun 202315 Jun 2023

Publication series

NameProceedings of the 24th Annual Conference of the European Association for Machine Translation, EAMT 2023

Conference

Conference24th Annual Conference of the European Association for Machine Translation, EAMT 2023
Country/TerritoryFinland
CityTampere
Period12/06/2315/06/23

ASJC Scopus subject areas

  • Language and Linguistics
  • Human-Computer Interaction
  • Software

Fingerprint

Dive into the research topics of 'Empirical Analysis of Beam Search Curse and Search Errors with Model Errors in Neural Machine Translation'. Together they form a unique fingerprint.

Cite this