A new double attention decoding model based on cascade RCNN and word embedding fusion for Chinese-English multimodal translation
by Haiying Liu
International Journal of Reasoning-based Intelligent Systems (IJRIS), Vol. 16, No. 1, 2024

Abstract: Traditional multimodal machine translation (MMT) is to optimise the translation process from the source language to the target language with the help of important feature information in images. However, the information in the image does not necessarily appear in the text, which will interfere with the translation. Compared with the reference translation, mistranslation can be appeared in the translation results. In order to solve above problems, we propose a double attention decoding method based on cascade RCNN to optimise existing multimodal neural machine translation models. The cascade RCNN is applied to source language and source image respectively. Word embedding is used to fuse the initialisation and the semantic information of the dual encoder. In attention computation process, it can reduce the focus on the repetitive information in the past. Finally, experiments are carried out on Chinese-English test sets to verify the effectiveness of the proposed method. Compared with other state-of-the-art methods, the proposed method can obtain better translation results.

Online publication date: Tue, 19-Mar-2024

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Reasoning-based Intelligent Systems (IJRIS):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com