欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
提出了一种基于单词分类的神经网络语言模型,以解决归一化问题。实验方法为,在基础翻译系统中加入模型参数,然后利用开发集调整参数,再对测试集进行翻译,对比加入模型参数前后的翻译质量以及训练模型和翻译过程所需时间。实验结果表明,在保证归一化的前提下,该模型的性能优于Vaswani等人的模型,且翻译质量与Vaswani等人的模型相当。
英文摘要：
A word classification-based neural network language model was proposed to resolve normalization problems. Model parameters were introduced to the basic translation system,which were adjusted by development sets. The test sets were translated. The translation quality and training model and the time taken by the translation were compared. The results indicate that the model is superior to that of Vasvani in performance with its translation quality being similar to that of Vasvani.
同期刊论文项目
同项目期刊论文
期刊信息
位置：