基于递归神经网络的英译汉机器翻译模型设计与实现

在自然语言处理领域,递归神经网络在机器翻译中的应用越来越广泛;除了其他语言外,汉语中还包含大量的词汇,提高英译汉的机器翻译质量是对汉语处理的一个重要贡献;设计了一个英汉机器翻译系统的模型,该系统使用基于知识的上下文向量来映射英语和汉语单词,采用编解码递归神经网络实现;对基于激活函数模型的性能进行了测试,测试结果表明,编码器层的线性激活函数和解码器层的双曲正切激活函数性能最好;从GRU和LSTM层的执行情况来看,GRU的性能优于LSTM;注意层采用softmax和sigmoid激活函数进行设置,该模型的方法在交叉熵损失度量方面优于现有的系统.

English to Chinese machine translation method based on recursive neural network

In the field of natural language processing, recursive neural networks are widely used in machine translation.In addition to other languages, Chinese contains a large number of words, and improving the quality of machine translation from English to Chinese is an important contribution to Chinese processing.This paper introduces the architecture of an English-Chinese machine translation system, which uses knowledge-based context vectors to map English and Chinese words, and is implemented by codec recursive neural network.This paper tests the performance based on the activation function model, and the test results show that the linear activation function of the encoder layer and the hyperbolic tangent activation function of the decoder layer have the best performance.From the execution of GRU and LSTM layer, GRU is better than LSTM.The attention layer is set by using Softmax and Sigmoid activation functions, and the method of this model is superior to existing systems in cross-entropy loss measurement.