Design of Bi LSTM Automatic Translation System Integrating transfer learning
In order to further improve the translation level of machine translation model from English to Chinese,an English Chi-nese automatic translation model is proposed based on transfer learning technology and two-way short-term memory network Bi LSTM.Among them,the basic Bi LSTM translation model is optimized through Gumbel Tree LSTM model,and then the idea of mi-gration pivot parameters in transfer learning is introduced to further optimize the model.The experimental results show that,compared with other translation models,the improved Bi LSTM English Chinese translation model GBi LSTM designed based on transfer learning has better translation quality,and the BLEU score and METEOR score on English French and English German corpus translation tests have reached 22.95％,36.02％,24.47％and 37.18％respectively;Compared with various baseline models,the introduction of transfer pivot model parameters significantly improved the translation quality of each translation model.The above results indicate that the designed GBi-LSTM translation model has excellent translation performance and can be applied to practical English Chinese trans-lation scenarios,with high feasibility.
万方数据
维普
