Improving Multilingual Sentence Embedding using Bi-directional Dual Encoder with Additive Margin Softmax
Abstract
In this paper, we present an approach to learn multilingual sentence embeddings using a bi-directional dual-encoder with additive margin softmax. The embeddings are able to achieve state-of-the-art results on the United Nations (UN) parallel corpus retrieval task. In all the languages tested, the system achieves P@1 of 86% or higher. We use pairs retrieved by our approach to train NMT models that achieve similar performance to models trained on gold pairs. We explore simple document-level embeddings constructed by averaging our sentence embeddings. On the UN document-level retrieval task, document embeddings achieve around 97% on P@1 for all experimented language pairs. Lastly, we evaluate the proposed model on the BUCC mining task. The learned embeddings with raw cosine similarity scores achieve competitive results compared to current state-of-the-art models, and with a second-stage scorer we achieve a new state-of-the-art level on this task.
Cite
@article{arxiv.1902.08564,
title = {Improving Multilingual Sentence Embedding using Bi-directional Dual Encoder with Additive Margin Softmax},
author = {Yinfei Yang and Gustavo Hernandez Abrego and Steve Yuan and Mandy Guo and Qinlan Shen and Daniel Cer and Yun-hsuan Sung and Brian Strope and Ray Kurzweil},
journal= {arXiv preprint arXiv:1902.08564},
year = {2019}
}
Comments
Accepted by IJCAI'19(International Joint Conference on Artificial Intelligence)