Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

GRU neural network Mongolian-Chinese machine translation method

A neural network and machine translation technology, applied in the field of machine translation, can solve problems such as unregistered word processing, low coding quality, missing translation, etc., and achieve the effects of reducing translation confusion, smooth expression ability, and performance improvement

Inactive Publication Date: 2020-01-31
INNER MONGOLIA UNIV OF TECH
View PDF5 Cites 11 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problems of missed translation, mistranslation, and unregistered word processing in the translation process mainly existing in the above-mentioned prior art, the purpose of the present invention is to provide a kind of GRU neural network Mongolian-Chinese machine translation method, which utilizes CPU and GPU to work in parallel Processing the corpus in this way nearly doubles the speed, and learning the corpus through the set learning rate can effectively alleviate the local optimal problem in the process of learning the semantic expression of the corpus and the problem of low coding quality caused by rapid convergence. , to improve the quality of the overall system by setting special structures and algorithms

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • GRU neural network Mongolian-Chinese machine translation method
  • GRU neural network Mongolian-Chinese machine translation method
  • GRU neural network Mongolian-Chinese machine translation method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0044] The implementation of the present invention will be described in detail below in conjunction with the drawings and examples.

[0045] The present invention is based on the GRU neural network, which is a variant of the LSTM neural network. LSTM neural network is proposed to overcome RNN's inability to handle long-distance dependencies well. The LSTM network is different from the traditional neural network, and the original neural network unit is transformed into a CEC memory unit. The summation mechanism of the CEC memory unit enables the gradient to be preserved and the error to be transmitted to solve the problem of gradient dispersion. The operation mechanism of LSTM is as follows: figure 1 .

[0046] The structure of the repeated network module of LSTM is much more complicated, and it realizes three gate calculations, namely forget gate, input gate and output gate. The forget gate is responsible for deciding how much to retain the unit state of the previous momen...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a GRU neural network Mongolian-Chinese machine translation method. Firstly, a translation language is preprocessed; then, an Encoder-Decoder model is built and trained for a certain scale of Mongolian and Chinese bilingual; the Mongolian and Chinese bilingual linguistic data is subjected to coding unified processing; and finally, a translation result is obtained on the basis of an Encoder-Decoder model, wherein the Encoder-Decoder model is constructed by a neural network, one neural network is LSTM; the encoder is responsible for Encoder encoding, bidirectional coding setting, namely, forward encoding and reverse encoding are carried out on a source language; the source statement is converted into two vectors which are coded in different directions and have fixed lengths, the other neural network is a GRU and is in charge of Decoder decoding; the decoding is carried out from a forward direction and a reverse direction, namely, the context information is automatically integrated when decoding is carried out to output the target language, thus the length-fixed vector generated by encoding is converted into the target sentence. According to the method, the characteristics of the Mongolian and Chinese language are combined, so that the expression capability of a Mongolian and Chinese machine translation system is smoother and closer to human expression, andthe degrees of semantic loss and translation disorder in the translation process are reduced.

Description

technical field [0001] The invention belongs to the technical field of machine translation, relates to Mongolian-Chinese machine translation, in particular to a Mongolian-Chinese machine translation method with a GRU neural network. Background technique [0002] At this stage, with the rapid development of the Internet industry and the continuous rise of a series of IT industries including information technology, machine translation for natural language processing is playing a certain role in promoting the development of the entire Internet industry. Large-scale search service industries such as Google and Baidu have conducted large-scale scientific research in the field of machine translation in the face of industry development. Continuous research in order to continuously obtain higher quality translations. [0003] Although scientific research institutions are still making continuous efforts to obtain better translation results, machine translation still exposes more and...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F40/58G06N3/08G06N3/04
CPCG06N3/084G06N3/044G06N3/045
Inventor 苏依拉卞乐乐赵旭薛媛范婷婷张振
Owner INNER MONGOLIA UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products