Center for Advanced Intelligence Project

The “RIKEN AIP Diversity Seminar” features outstanding female researchers in AI delivering lectures in a hybrid format.

This talk will be held in a hybrid format, both in person at AIP Open Space of RIKEN AIP (Nihonbashi office) and online by Zoom. AIP Open Space: *only available to AIP researchers.

Date & Time:
Feb. 26, 2025 : 13:30 pm – 14:30 pm (JST)

Title:
Bridging Linguistic Gaps to Enhance Multilingual Translation

Speaker:
Dr. Akiko Eriguchi (Senior Researcher, Microsoft)

Abstract:
In this talk, I will present our recent research work aimed at improving multilingual capabilities such as translation using neural machine translation and large language models. The focus will be on how to enable one model to support a wide range of languages at scale. The discussion will emphasize how these techniques address the challenges of translating mid- and low-resource languages and adapting vocabularies of new languages with different scripts. Additionally, I will highlight our proposed methods 1) to ensure high-quality translation across 50 diverse languages, including the use of plug-and-play language-specific modules and adaptive rejection optimization, and 2) to adapt new languages using adaptor networks. These methods not only enhance the performance of multilingual models but also pave the way for higher linguistic inclusiveness. By leveraging the approaches, we aim to bridge the gap between high-resource and low-resource languages, ultimately contributing to a better connected and linguistically diverse world.

Biography:
Akiko Eriguchi, Ph.D. Akiko is a Senior Researcher at Microsoft. Her research interests lie in multilingual natural language processing (NLP) and machine learning. With her position in the Microsoft Translator team, she has played a pivotal role in developing machine translation systems and multilingual NLP applications. Akiko is also deeply committed to community engagement, serving on the Board of Directors at Women in Machine Learning, Inc., where she works to empower women and/or non-binary people in the field of machine learning.
Prior to joining Microsoft, she was a Research Fellow (DC1) at the Japan Society for the Promotion of Science. She obtained her doctoral degree from the University of Tokyo, Japan, where her PhD thesis received the sixth AAMT Nagao Student award from the Asia-Pacific Association for Machine Translation.