Aligned Cross Entropy for Non-Autoregressive Machine Translation

被引:0
|
作者
Ghazvininejad, Marjan [1 ]
Karpukhin, Vladimir [1 ]
Zettlemoyer, Luke [1 ]
Levy, Omer [1 ]
机构
[1] Facebook AI Res, Menlo Pk, CA 94025 USA
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Non-autoregressive machine translation models significantly speed up decoding by allowing for parallel prediction of the entire target sequence. However, modeling word order is more challenging due to the lack of autoregressive factors in the model. This difficultly is compounded during training with cross entropy loss, which can highly penalize small shifts in word order. In this paper, we propose aligned cross entropy (AXE) as an alternative loss function for training of non-autoregressive models. AXE uses a differentiable dynamic program to assign loss based on the best possible monotonic alignment between target tokens and model predictions. AXE-based training of conditional masked language models (CMLMs) substantially improves performance on major WMT benchmarks, while setting a new state of the art for non-autoregressive models.
引用
收藏
页数:9
相关论文
共 50 条
  • [1] Aligned Cross Entropy for Non-Autoregressive Machine Translation
    Ghazvininejad, Marjan
    Karpukhin, Vladimir
    Zettlemoyer, Luke
    Levy, Omer
    25TH AMERICAS CONFERENCE ON INFORMATION SYSTEMS (AMCIS 2019), 2019,
  • [2] Order-Agnostic Cross Entropy for Non-Autoregressive Machine Translation
    Du, Cunxiao
    Tu, Zhaopeng
    Jiang, Jing
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 139, 2021, 139
  • [3] Integrating Translation Memories into Non-Autoregressive Machine Translation
    Xu, Jitao
    Crego, Josep
    Yvon, Francois
    17TH CONFERENCE OF THE EUROPEAN CHAPTER OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS, EACL 2023, 2023, : 1326 - 1338
  • [4] Enhanced encoder for non-autoregressive machine translation
    Wang, Shuheng
    Shi, Shumin
    Huang, Heyan
    MACHINE TRANSLATION, 2021, 35 (04) : 595 - 609
  • [5] Rephrasing the Reference for Non-autoregressive Machine Translation
    Shao, Chenze
    Zhang, Jinchao
    Zhou, Jie
    Feng, Yang
    THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 11, 2023, : 13538 - 13546
  • [6] Acyclic Transformer for Non-Autoregressive Machine Translation
    Huang, Fei
    Zhou, Hao
    Liu, Yang
    Li, Hang
    Huang, Minlie
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 162, 2022,
  • [7] Non-Autoregressive Machine Translation with Auxiliary Regularization
    Wang, Yiren
    Tian, Fei
    He, Di
    Qin, Tao
    Zhai, ChengXiang
    Liu, Tie-Yan
    THIRTY-THIRD AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTY-FIRST INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE / NINTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2019, : 5377 - 5384
  • [8] A Survey of Non-Autoregressive Neural Machine Translation
    Li, Feng
    Chen, Jingxian
    Zhang, Xuejun
    ELECTRONICS, 2023, 12 (13)
  • [9] Non-Autoregressive Machine Translation as Constrained HMM
    Li, Haoran
    Jie, Zhanming
    Lui, Wei
    FINDINGS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS: ACL 2024, 2024, : 12361 - 12372
  • [10] Non-Autoregressive Machine Translation with Latent Alignments
    Saharia, Chitwan
    Chan, William
    Saxena, Saurabh
    Norouzi, Mohammad
    PROCEEDINGS OF THE 2020 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING (EMNLP), 2020, : 1098 - 1108