November 7, 2019 - November 8, 2019

MSRA Academic Day 2019

Location: Beijing, China

November 7

  • Abstract: We live in a world of connected entities including various systems (ranging from big cloud and edge systems to individual memory and disk systems) networked together. Innovations in systems and networking are key driving forces in the era of big data and artificial intelligence, to empower advanced intelligent algorithms with reliable, secure, scalable and efficient computing capacity to process huge volumes of data. We have witnessed the significant progress in cloud systems, and recently, edge computing, in particular AI on Edge, has attracted increasing attention from both academia and industry. This workshop aims to report and discuss the most recent progress and trends on general system and networking area, especially on various infrastructure support for machine learning systems.

    Event owners: Yunxin Liu, Yongqiang Xiong

    Time (CST) Workshops Speaker Location
    2:00 PM–2:10 PM Welcome and introductions Yunxin Liu & Yongqiang Xiong, Microsoft Research Dong Zhi Men, Microsoft Tower 1-1F
    2:10 PM–3:25 PM Research at Microsoft (25 mins per talk x3)
    • Peng Cheng, Microsoft Research
    • Ting Cao, Microsoft Research
    • Quanlu Zhang, Microsoft Research
    3:25 PM–4:40 PM Research talks (25 mins per talk x3)
    • Chuan Wu, University of Hong Kong
    • Xuanzhe Liu, Peking University
    • Rajesh Krishna Balan, Singapore Management University
    4:40 PM–5:20 PM Panel with discussion

    Title: “What’s missing in system & networking for AI?”
    • Yunxin Liu, Microsoft Research (Moderator)
    • Yongqiang Xiong, Microsoft Research (Moderator)
    • Chuan Wu, University of Hong Kong
    • Xuanzhe Liu, Peking University
    • Rajesh Krishna Balan, Singapore Management University
    • Peng Cheng, Microsoft Research
    • Ting Cao, Microsoft Research
    • Quanlu Zhang, Microsoft Research
    5:20 PM–5:30 PM Wrap-up and closing

  • Abstract: Deep learning has greatly driven this wave of AI. While deep learning has made many breakthroughs in recent years, its success heavily relies on big labeled data, big model, and big computing. As edge computing becomes the trend and more and more IoT devices become available, deep learning faces the low-resource challenge: how to learn from limited labeled data, with limited model size, and limited computation resources. The theme of this workshop is low-resource machine learning: learning from low-resource data, learning compact models, and learning with limited computational resources. This workshop aims to report latest progress and discuss the trends and frontiers of research on low-resource machine learning.

    Event owner: Tao Qin

    Time (CST) Workshops Speaker Location
    2:00 PM–2:10 PM Welcome and introductions Tao Qin, Microsoft Research Xi Zhi Men, Microsoft Tower 1-1F
    2:10 PM–3:25 PM Research at Microsoft (25 mins per talk x3)
    • Yingce Xia, Microsoft Research
    • Xu Tan, Microsoft Research
    • Guolin Ke, Microsoft Research
    3:25 PM–4:40 PM Research talks (25 mins per talk x3)
    • Jaegul Choo, Korea University
    • Sinno Jialin Pan, Nanyang Technological University
    • Sung Ju Hwang, KAIST
    4:40 PM–5:20 PM Panel with discussion

    Title: “Challenges and Future of Low-Resource Machine Leaning”
    • Tao Qin, Microsoft Research (Moderator)
    • Jaegul Choo, Korea University
    • Sung Ju Hwang, KAIST
    • Shujie Liu, Microsoft Research
    • Dongdong Zhang, Microsoft Research
    5:20 PM–5:30 PM Wrap-up and closing

  • Abstract: We live in a world of multimedia (text, image, video, audio, sensor data, 3D, etc.). These modalities are integral components of real-world events and applications. A full understanding of multimedia relies heavily on feature learning, entity recognition, knowledge, reasoning, language representation, etc. Cross-modal learning, which requires joint feature learning and cross-modal relationship modeling, has attracted increasing attention from both academia and industry. This workshop aims to report and discuss the most recent progress and trends on multimodal representation learning for multimedia applications.

    Event owners: Wenjun Zeng, Nan Duan

    Time (CST) Workshops Speaker Location
    2:00 PM–2:10 PM Welcome and introductions Wenjun Zeng, Microsoft Research Tian An Men, Microsoft Tower 1-1F
    2:10 PM–3:25 PM Research at Microsoft (25 mins per talk x3)
    • Nan Duan, Microsoft Research
    • Yue Cao, Microsoft Research
    • Chong Luo, Microsoft Research
    3:25 PM–4:40 PM Research talks (25 mins per talk x3)
    • Gunhee Kim, Seoul National University
    • Winston Hsu, National Taiwan University
    • Jiwen Lu, Tsinghua University
    4:40 PM–5:20 PM Panel with discussion

    Title: Opportunities and Challenges for Cross-Modal Learning
    • Wenjun Zeng, Microsoft Research (Moderator)
    • Xilin Chen, Chinese Academy of Science
    • Winston Hsu, National Taiwan University
    • Gunhee Kim, Seoul National University
    • Nan Duan, Microsoft Research
    5:20 PM–5:30 PM Wrap-up and closing

November 8

Time (CST) Workshops Speaker Location
09:00 – 09:30 Welcome & MSRA Overview Hsiao-Wuen Hon Gu Gong, Microsoft Tower 1-1F
09:30 – 09:40 Fellowship Award Ceremony Presenter: Hsiao-Wuen Hon
09:40 – 10:00 Photo session & Break
10:00 – 10:40 Panel Discussion

Title: “How to foster a computer scientist”

 Moderator: Tim Pan, Microsoft Research

Panelists:

  • Bohyung Han, Seoul National University
  • Junichi Rekimoto, The University of Tokyo
  • Winston Hsu, National Taiwan University
  • Xin Tong, Microsoft Research
10:40 – 11:55 Technology Showcase by Microsoft Research Asia (5)
  • “OneOCR For Digital Transformation” by Qiang Huo
  • “NN grammar check” by Tao Ge
  • “AutoSys: Learning based approach for system optimization” by Mao Yang
  • “Dual learning and its application in translation and speech from ML” by Tao Qin(Yingce Xia and Xu Tan)
  • “Spreadsheet Intelligence for Ideas in Excel” by Shi Han
12:00 – 14:00 Technology Showcase by Academic Collaborators Lunch, Microsoft Tower1-1F
14:00 – 17:30 Breakout Sessions
Language and Knowledge Leader: Xing Xie

Speakers: Seung-won Hwang, Min Zhang, Lei Chen, Masatoshi Yoshikawa, Shou-De Lin, Rui Yan, Hiroaki Yamane, Chenhui Chu, Tadashi Nomoto

 Zhong Guan Cun, Microsoft Tower 2-4F
System and Networking Leaders: Lidong Zhou, Yunxin Liu

Speakers: Insik Shin, Wenfei Wu, Rajesh Krishna Balan, Youyou Lu, Chuck Yoo, Yu Zhang, Atsuko Miyaji, Jingwen Leng, Yao Guo, Heejo Lee, Cheng Li

 San Li Tun, Microsoft Tower 2-4F
Computer Vision Leader: Wenjun Zeng

Speakers: Gunhee Kim, Tianzhu Zhang, Yonggang Wen, Wen-Huang Cheng, Jiaying Liu, Bohyung Han, Wei-Shi Zheng, Jun Takamatsu, Xueming Qian

 Qian Men, Microsoft Tower 2-4F
Graphics Leader: Xin Tong

Speakers: Min H. Kim, Seungyong Lee, Sung-eui Yoon

 Di Tan, Microsoft Tower 2-4F
Multimedia Leader: Yan Lu

Speakers: Seung Ah Lee, Huanjing Yue, Hiroki Watanabe, Minsu Cho, Zhou Zhao, Seungmoon Choi

 Gu Lou, Microsoft Tower 2-4F
Healthcare Leader: Eric Chang

Speakers: Ryo Furukawa, Winston Hsu

 Dong Cheng, Microsoft Tower 2-4F
Data, Knowledge, and Intelligence Leaders: Jian-Guang Lou, Qingwei Lin

Speakers: Shixia Liu, Huamin Qu, Jong Kim, Yingcai Wu

 Xi Cheng, Microsoft Tower 2-4F
Machine Learning Leader: Tao Qin

Speakers: Hongzhi Wang, Seong-Whan Lee, Sinno Jialin Pan, Lijun Zhang, Jaegul Choo, Mingkui Tan, Liwei Wang

 Ri Tan, Microsoft Tower 2-4F
Speech Leader: Frank Soong

Speakers: Jun Du, Hong-Goo Kang

 Guo Zi Jian, Microsoft Tower 2-4F
17:30-18:00 Transition Break
18:15 – 20:30 Banquet Ballroom located @ 3F, Tylfull Hotel