RNN编码器-解码器在维汉机器翻译中的应用
|
摘要
将RNN编码器-解码器作为传统的基于短语的PSMT系统的一部分,在传统统计机器翻译基础上,集成RNN解码器-编码器,兼容PSMT创建了新联合模型(RNN+PSMT)。新的模型不仅在维-汉、汉-英机器翻译的应用中取得了成效,而且能够捕捉到语言的规律,使得机器翻译中的一个重要评价指标的BLEU值得到了显著提高。实验结果表明,系统的整体性能超过了传统统计机器翻译。
In this paper, the RNN encoder-decoder is used as part of the traditional phrase-based PSMT system. On the basis of traditional statistical machine translation, the integrated RNN encoder-decoder, compatible with PSMT(Phrase based Statistical Machine Translation)has created a new Uyghur-Chinese neural machine translation model. The new model not only has achieved good results in the application of Uyghur-Chinese and Chinese-English machine translation, but also can capture the rule of language, which makes the BLEU value of an important evaluation indicator greatly improved in Machine Translation. Experimental results show that the overall performance of the system exceeds the traditional statistical machine translation.
In this paper, the RNN encoder-decoder is used as part of the traditional phrase-based PSMT system. On the basis of traditional statistical machine translation, the integrated RNN encoder-decoder, compatible with PSMT(Phrase based Statistical Machine Translation)has created a new Uyghur-Chinese neural machine translation model. The new model not only has achieved good results in the application of Uyghur-Chinese and Chinese-English machine translation, but also can capture the rule of language, which makes the BLEU value of an important evaluation indicator greatly improved in Machine Translation. Experimental results show that the overall performance of the system exceeds the traditional statistical machine translation.
引文
[1]Forcada M L,?eco R P.Recursive hetero-associative memories for translation[C]//International Work-Conference on Artificial and Natural Neural Networks:Biological and Artificial Computation:From Neuroscience To Technology.[S.l.]:Springer-Verlag,1997.
[2]Kalchbrenner N,Blunsom P.Recurrent continuous translation models[C]//Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing,2013.
[3]百度.人工智能领域新突破百度发布基于深度学习的在线翻译系统[EB/OL].(2015).http://news.xinhuanet.com/2015-05/21/c_1115367884.htm.
[4]远洋.谷歌启用神经机器翻译系统[EB/OL].(2016).https://www.ithome.com/html/it/273102.htm.
[5]王丽.双语词典在统计机器翻译中的应用[D].哈尔滨:黑龙江大学,2009.
[6]陈韵,张鹏华,任利华.机器翻译研究述评[D].西安:西安电子科技大学,2013.
[7]Hochreiter S,Schmidhuber J.Long short-term memory[J].Neural Computation,1997,9(8):1735-1780.
[8]Chung J,Gulcehre C,Cho K,et al.Empirical evaluation of gated recurrent neural networks on sequence modeling[EB/OL].(2014-12-11).https://arxiv.org/abs/1412.3555.
[9]Bougares F,Schwenk H,Bengio Y.Learning phrase representations using RNN encoder-decoder for statistical machine translation[J].Computer Science,2014.
[10]Auli M,Galley M,Quirk C,et al.Joint language and translation modeling with recurrent neural networks[C]//Proceedings of the ACL Conference on Empirical Methods in Natural Language Processing(EMNLP),2013:1044-1054.
[11]Axelrod A,He Xiaodong,Gao Jianfeng.Domain adaptation via pseudo in-domain data selection[C]//Proceedings of the ACL Conference on Empirical Methods in Natural Language Processing(EMNLP),2011:355-362.
[12]Bastien F,Lamblin P,Pascanu R,et al.Theano:New features and speed improvements[C]//2012 Workshop on Deep Learning and Unsupervised Feature Learning,2012.
[13]Bahdanau D,Cho K,Bengio Y.Neural machine translation by jointly learning to align and translate[C]//Proceedings of ICLR,2015.
[2]Kalchbrenner N,Blunsom P.Recurrent continuous translation models[C]//Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing,2013.
[3]百度.人工智能领域新突破百度发布基于深度学习的在线翻译系统[EB/OL].(2015).http://news.xinhuanet.com/2015-05/21/c_1115367884.htm.
[4]远洋.谷歌启用神经机器翻译系统[EB/OL].(2016).https://www.ithome.com/html/it/273102.htm.
[5]王丽.双语词典在统计机器翻译中的应用[D].哈尔滨:黑龙江大学,2009.
[6]陈韵,张鹏华,任利华.机器翻译研究述评[D].西安:西安电子科技大学,2013.
[7]Hochreiter S,Schmidhuber J.Long short-term memory[J].Neural Computation,1997,9(8):1735-1780.
[8]Chung J,Gulcehre C,Cho K,et al.Empirical evaluation of gated recurrent neural networks on sequence modeling[EB/OL].(2014-12-11).https://arxiv.org/abs/1412.3555.
[9]Bougares F,Schwenk H,Bengio Y.Learning phrase representations using RNN encoder-decoder for statistical machine translation[J].Computer Science,2014.
[10]Auli M,Galley M,Quirk C,et al.Joint language and translation modeling with recurrent neural networks[C]//Proceedings of the ACL Conference on Empirical Methods in Natural Language Processing(EMNLP),2013:1044-1054.
[11]Axelrod A,He Xiaodong,Gao Jianfeng.Domain adaptation via pseudo in-domain data selection[C]//Proceedings of the ACL Conference on Empirical Methods in Natural Language Processing(EMNLP),2011:355-362.
[12]Bastien F,Lamblin P,Pascanu R,et al.Theano:New features and speed improvements[C]//2012 Workshop on Deep Learning and Unsupervised Feature Learning,2012.
[13]Bahdanau D,Cho K,Bengio Y.Neural machine translation by jointly learning to align and translate[C]//Proceedings of ICLR,2015.