Return to Microsoft Research Lab - Asia

Deep and Reinforcement Learning Group

The Deep and Reinforcement Learning Group at Microsoft Research Asia pushes forward the research of deep learning and reinforcement learning from both algorithmic and practical aspects. Our interests include

 

Our dual learning (opens in new tab) and other techniques helped Microsoft achieve human parity in Chinese-English (opens in new tab) news translation in 2018, and win the first place for 8 translation tasks (opens in new tab) in WMT 2019. We built the world-best Mahjong AI (opens in new tab), named Suphx (opens in new tab), which achieved 10 DAN on the Tenhou platform in mid 2019. Our FastSpeech (opens in new tab) model is the backbone of Azure TTS (opens in new tab) and has supported 50+ languages/locals and 80+ voices.

People

Portrait of Tao Qin

Tao Qin

Senior Principal Research Manager

Portrait of Li Zhao

Li Zhao

Principal Researcher

Portrait of Xu Tan

Xu Tan

Principal Research Manager

Portrait of Yingce Xia

Yingce Xia

Principle Researcher

Portrait of Shufang Xie

Shufang Xie

SENIOR RESEARCH SDE

Portrait of Rui Wang

Rui Wang

Senior Researcher

Portrait of Jindong Wang

Jindong Wang

Senior Researcher

Portrait of Chang Liu

Chang Liu

Senior Researcher

Portrait of Renqian Luo

Renqian Luo

Senior Researcher

Portrait of Guoqing Liu

Guoqing Liu

Senior Researcher

Portrait of Junliang Guo

Junliang Guo

Researcher