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

Neural Machine Translation Method Based on Rereading and Feedback Mechanism

A feedback mechanism and machine translation technology, applied in neural architecture, natural language translation, biological neural network models, etc., can solve problems such as insufficient utilization of global information

Active Publication Date: 2021-05-11
KUNMING UNIV OF SCI & TECH
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a neural machine translation method based on rereading and feedback mechanism to solve the problem of insufficient utilization of global information in a parallel corpus scarcity environment

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
  • Neural Machine Translation Method Based on Rereading and Feedback Mechanism
  • Neural Machine Translation Method Based on Rereading and Feedback Mechanism
  • Neural Machine Translation Method Based on Rereading and Feedback Mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] Embodiment 1: as Figure 1-5 As shown, the neural machine translation method based on the rereading and feedback mechanism, the specific steps are as follows:

[0041] Step1. Corpus preprocessing: Preprocess parallel training corpus, verification corpus and test corpus of different scales for model training, parameter tuning and effect testing. The source language training corpus is marked as x;

[0042] The processed parallel corpus is divided into three categories according to the scale: small scale, medium scale and large scale. By applying the method of the present invention to parallel corpora of different scales, the influence of the increase in corpus scale on information utilization can be observed, and the hypothesis that the proposed method is applicable to scenarios where parallel corpus resources are scarce can be verified. Table 1 is the experimental data information.

[0043] Table 1 Experimental data

[0044]

[0045] Step2, the first layer of encod...

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 relates to a neural machine translation method based on a rereading and feedback mechanism, and belongs to the technical field of natural language processing. First, the final state of the first-layer encoder is passed to the first-layer decoder for decoding, and its copy is generated and passed to the second-layer encoder. The second-layer encoder re-reads the state for parameter initialization. This process is called "rereading". Secondly, the decoding result of the first-layer decoder and the source language are used as the input of the importance calculation method, and the generated importance weights are fed back to the second-layer encoder to guide the encoding. This process is called "feedback". The invention can realize the change of the end-to-end neural network structure, and the changed network structure can better mine the global information, and is suitable for the parallel corpus-scarce translation environment.

Description

technical field [0001] The invention relates to a neural machine translation method based on a rereading and feedback mechanism, and belongs to the technical field of natural language processing. Background technique [0002] Neural machine translation is essentially a data-driven language conversion task, and the data that has an important impact on its performance is a parallel corpus. In scenarios with abundant parallel corpus resources (such as English-French, Chinese-English, etc.), neural machine translation has surpassed traditional statistical machine translation in terms of performance. However, in scenarios where parallel corpus resources are scarce (such as Chinese-Southeast Asian), the performance of neural machine translation is not very ideal. Therefore, exploring how to mine more information from limited parallel corpus has very important research and application value. [0003] At present, in mining potential information in the corpus by changing the neural...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G06N3/04G06F40/42G06F40/30G06F40/211
CPCG06N3/044
Inventor 余正涛于志强郭军军文永华高盛祥王振晗
Owner KUNMING UNIV OF SCI & 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

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com