Lei Li

Research

• Machine translation, speech translation, multilingual NLP.
• Text generation and summarization.
• Reasoning and question answering.
• Information extraction.
• AI for drug discovery
• Green and Efficient ML
• Time series mining and prediction
• Probabilistic inference, Bayesian sampling methods

Affiliations

UCSB NLP Group

Biography

Lei Li is an assistant professor in Computer Science Department at University of California Santa Barbara. His research interest lies in natural language pro-cessing, machine translation, and machine learning. He received his B.S. from Shanghai Jiao Tong University and Ph.D. from Carnegie Mellon University. His dissertation work on fast algorithms for mining co-evolving time series was awarded ACM KDD best dissertation (runner up). His recent work on AI writer Xiaomingbot received 2nd-class award of Wu Wen-tsün AI prize in 2017. He is a recipient of ACL 2021 best paper award, CCF Young Elite award in 2019, and CCF distinguished speaker in 2017. His team won first places for five language translation directions and the best in corpus filtering challenge in WMT 2020. Previously, he worked as a post-doctoral researcher at EECS department of UC Berkeley, a principal researcher at Baidu's Institute of Deep Learning in Silicon Valley, and at ByteDance as the founding director of AI Lab. He has served as Associate Editor of TPAMI and organizers and area chair/senior PC for multiple conferences including ACL, EMNLP, NeurIPS, KDD, AAAI, IJCAI, WSDM, and CIKM. He has launched ByteDance's machine translation system (VolcTrans) and Xiaomingbot automatic writing system, and many of his algorithms have been deployed in production (Toutiao, Douyin, Tiktok, Xigua, Feishu/Lark), serving over a billion users. He has delivered five tutorials at ACL 2021, EMNLP 2019, NLPCC 2019, NLPCC 2016, and KDD 2010. He was a lecturer for 2014 Probabilistic Programming for Advancing Machine Learning summer school at Portland, USA.

Honors

1. ACL 2021 Best Paper Award (1 in 3350 submissions).
2. Winning 1st places in WMT20 Machine Translation Contest for 5 language directions: Chinese-English,
German-English, French-German, English-Khmer, and English-Pashto, as well as 1st places in Parallel
Corpus Filtering competition for Khmer and Pashto. November, 2020.
3. 2019 Young Elite Award by China Computer Federation. (6 winners in total)
4. 2017 2nd-class Wu Wen-tsün Award in AI for “Key technologies of Internet information summarization
and machine writing”, the premier award by China Association of Artificial Intelligence.
5. 2017 China Computer Federation Distinguished Speaker.
6. 2012 SIGKDD Doctoral Dissertation Award (runner-up), the Association for Computing Machinery (3
winners and runner-ups in total).
7. National Excellent Patent award in 2021, China, for patent 201811060837.X.
8. Led the development of VolcTrans, a machine translation system, with billions of requests daily, serving
over a billion of active users.
9. Led the development of AI writing bot Xiaomingbot.
• It automatically generates sports news about 17 leagues/series including the Rio Olympics (2016)
and 2018 FIFA World Cup, with over 150,000 followers on social media. The work attracted over
100 coverage from news outlets including MIT Technology Review, Newsweek, ChinaDaily, and Y
Combinator.
• It is also deployed in an online advertisement platform, which helps hundreds of thousands of small
and mid-sized enterprises to create better advertisements.
• Xiaomingbot also automatically generates timely missing-person announcements in text and videos
for the Toutiao alert program of finding missing people with crowdsourcing, leading to more than
10,000 missing people found during 2016-2019. Multiple news coverage including GlobalTimes and
Xinhuanet.
• It automatically writes product promotion articles in the Poverty-Alleviation program to help farm-
ers to boost agriculture product sales on Toutiao platform, helping 20,000 families. Multiple news
coverage including Xinhuanet.
10. Led the development of a large scale video duplication and infringement detection system, serving over
one billion active users and analyzing 20 million videos daily by the end of 2019.
11. Led the development of key technologies of query analysis for Toutiao Search engine, which serves
hundreds of millions of users daily, bringing in significant revenue.

Education

PhD  Carnegie Mellon University
BS     Shanghai Jiao Tong University