Aligning Multilingual Word Embeddings for Cross-Modal Retrieval Task
TL
TL;DR
It is shown that the proposed new approach to learn multimodal multilingual embeddings for matching images and their relevant captions in two languages enables better generalization, achieving state-of-the-art performance in text-to-image and image- to-text retrieval task, and caption-caption similarity task.
Abstract
In this paper, we propose a new approach to learn multimodal multilingual embeddings for matching images and their relevant captions in two languages. We combine two existing objective functions to make images and captions close in a joint embedding space while adapting the alignment of word embeddings between existing languages in our model. We show that our approach enables better generalization, achieving state-of-the-art performance in text-to-image and image-to-text retrieval task, and caption-caption similarity task. Two multimodal multilingual datasets are used for evaluation: Multi30k with German and English captions and Microsoft-COCO with English and Japanese captions.
References40 items
1
Loss in Translation: Learning Bilingual Word Mapping with a Retrieval Criterion
2
Neural Baby Talk
3
Unsupervised Machine Translation Using Monolingual Corpora Only
4
Word Translation Without Parallel Data
5