• Home
  • News
  • [Matsuo-Lab. University of Tokyo] Report on 2020 Activities
  • [Matsuo-Lab. University of Tokyo] Report on 2020 Activities

    Hello, this is the Public Relations Team of the Matsuo Institute.

    The turbulent year 2020 will soon be over.
    For Matsuo Lab, 2020 was a year of many positive changes.

    Today, we reiterate some of the topics for 2020 that have been discussed both inside and outside the lab!

    1. Strengthen basic research team

    Three new basic research members joined Matsuo Lab this year.

    The organization became more and more lively with the start of exchanges with Dr. Hiroshi Maruyama, who was appointed as a specially-appointed professor at the Artificial Engineering Research Center, to which Dr. Matsuo belongs, and the join-in of three robots.

    In August, Masahiro Suzuki, a specially-appointed researcher, was appointed as a specially-appointed assistant professor, and in December, Yusuke Iwasawa, a specially-appointed assistant professor, was appointed as a specially-appointed lecturer!

    Introduction of New Research Members

    Project Assistant Professor Wataru Kumagai

    Profile
    2013 Completed Doctoral program, Graduate School of Information Science and Technology, Tohoku University (JSPS DC1)
    Sep. 2013-Mar. 2014 JSPS PD, Nagoya University
    Apr 2014-Mar 2017 Specially Appointed Assistant Professor, Faculty of Engineering, Kanagawa University
    Apr. 2017-Jul. 2020 Researcher, RIKEN Center for Integrated Research on Innovative Intelligence
    Aug. 2020- Current position

    Project Assistant Professor Keiichi Ochiai

    Profile
    March 2008 Completed Master’s degree in Intelligent Information Engineering, Graduate School of Natural Science and Technology, Chiba University Master’s degree (Engineering)
    April 2008 Joined NTT DOCOMO, INC.
    September 2017 Completed Doctoral program in Technology Management Strategy, Graduate School of Engineering, The University of Tokyo Doctor (Engineering)
    August 2020 Specially Appointed Assistant Professor, Matsuo Laboratory, The University of Tokyo
    KDD CUP 2019 Regular ML Track 1st Prize, ICWSM2020 Best Paper Honorable Mentions

    Ms. Machelle O’Neill, Research Engineer Lead

    Profile.
    2019-2020 Research Intern, The University of Tokyo
    2020 Student, Masayoshi Son Scholarship Foundation
    2020 Researcher, The University of Tokyo

    Dr. Maruyama becomes a specially-appointed professor at the Artificial Engineering Research Center and begins exchanges with the Matsuo Institute.

    Dr. Hiroshi Maruyama, Specially Appointed Professor, Artificial Engineering Research Center

    Profile
    1983 M.S. Tokyo Institute of Technology
    Joined IBM Japan in the same year
    Engaged in research on artificial intelligence and natural language processing at Japan Science Institute (later Tokyo Research Laboratory)
    1997-2000 Visiting Associate Professor, Graduate School of Information Science and Technology, Tokyo Institute of Technology: Research, development, and standardization of XML, Web services, and security
    2003-2004 Seconded to IBM Business Consulting Services, Inc.
    2006-2009 Director, Tokyo Research Laboratory Executive Officer
    2009-2010 Deputy General Manager, Digital Platform Development Division, Canon Inc.
    2011-2016 Professor, The Institute of Statistical Mathematics, Research Organization of Information and Systems
    2016-2018 Chief Strategy Officer, Preferred Networks, Inc.
    April 2018-Present PFN Fellow, Preferred Networks, Inc.

    2. Papers

    Matsuo Lab’s research covers a wide range of areas in machine learning and deep learning.
    This year, our work has been accepted for publication in the following international conferences and journals.

    ◆IJCAI2020

    Stabilizing Adversarial Invariance Induction from a Divergence Minimization Perspective
     Author: Yusuke Iwasawa, Kei Akuzawa, Yutaka Matsuo

    ◆Transactions of the Japanese Society for Artificial Intelligence

    Modeling and Analysis of the Effects on Financial Markets from Influence Relations in Social Media
     Author: Koichiro Tamura, Yutaka Matsuo

    ◆EMNLP2020

    VCDM: Leveraging Variational Bi-encoding and Deep Contextualized Word Representations for Improved Definition Modeling
     Authors: Machel Reid, Edison Marrese-Taylor, Yutaka Matsuo

    Frontiers in Robotics & AI

    Modeling Task Uncertainty for Safe Meta-Imitation Learning
     Author: Tatsuya Matsushima, Naruya Kondo, Yusuke Iwasawa, Kaoru Nasuno, Yutaka Matsuo

    ◆AAAI2021

    Variational Inference for Learning Representations of Natural Language Edits
     Author: Edison Marrese-Taylor, Machel Reid, Yutaka Matsuo

    3. Lecture

    Matsuo Lab offers various lectures on deep learning, data science, and web engineering.
    A total of more than 2,000 people attend the courses each year. (The photo shows an offline session held in 2019 or earlier.)

    ◆ “Reinforcement Learning” course started at ◆ Summer School, three-course system

    For students who have already mastered the basics of deep learning, we offer short-term intensive courses during the spring and summer vacations as an opportunity to learn intensively about a specific area.

    In recent years, only two courses, “Natural Language Processing” and “Generative Modeling,” were offered, but this August, a “Reinforcement Learning” course was also started.

    ☆ March 2021: Spring Seminar begins.
    Information will be released around January 2021.
    We will announce on Twitter (@Matsuo_Lab) when we start accepting applications.

    ◆83 technical college students participated in “AI Summer School” hosted by MAiZM

    The “AI Summer School” for technical college students, organized by the Mitoyo AI Society Promotion Organization (MAiZM), which was launched by Mitoyo City, Matsuo Lab, and Kagawa National College of Technology last April 2019, opened again this year following last year. Members of the Matsuo Lab gave lectures.

    Since the event was held online this year, we widely invited participation of technical college students from all over Japan in addition to Kagawa National College of Technology, and 83 students participated in the event.

    4. Various study groups and circle activities

    Matsuo Lab conducts various study groups and circle activities for researchers and students to implement the latest papers.

    ◆256th DL reading session is held.

    In December, the DL reading group, in which participants read and presented the latest papers on Deep Learning, held its 256th meeting (2 to the power of 8). On the day of the 256th meeting, a special lecture and reception were held, and more than 70 people, including alumni who had not been to the meeting for a while, registered for the event. (Slide photos from the day are shown below.)

    ◆Robocon team launched andawarded at RoboCup Japan Open 2020

    A new robotics team has been formed with the goal of winning the WRS (World Robot Summit 2020), which has been postponed until next year.

    We also participated in the RoboCup Japan Open 2020 in December, where we placed second in the @Home Domestic Standard Platform League (DSPL) and first in the Technical Challenge!

    YouTube distribution: https://youtu.be/GYFba6NVkrs

    5. media coverage

    This year, many media outlets introduced the activities of the Matsuo Institute.
    The Matsuo Institute was also featured in an international journal of science and technology, providing an opportunity for the Institute to be introduced to the world.

    ◆Matsuo Lab. was featured in “Nature Index” AI special issue.

    Matsuo Lab’s research, industry-academia collaboration, and entrepreneurship development efforts were introduced in the Nature Index AI feature article published in December.

    Raising a global center for deep learning

    6. Industry-University Collaboration

    ◆Dr. Matsuo becomes Technical Advisor to Matsuo Research Institute, Inc.

    Matsuo Institute, Inc. was established in February 2020 to create and develop an “ecosystem” of innovation centered at the university, accompanied by the Matsuo Laboratory of the University of Tokyo.

    ▪ Matsuo Research Institute, Inc. Corporate Charter
    http://matsuo-institute.com/charter/

    Matsuo Institute, Inc., with Silicon Valley as one of its benchmarks, is committed to contributing to industry through social implementation of advanced technologies and fostering AI technology developers and entrepreneurs, and to giving back to the research activities of the next generation.
    We aim to become a role model of industry-academia collaboration and the starting point of an innovation spiral.

    Please refer to this interview article to learn more about how Matsuo Institute, Inc. was established.

    https://www.bizreach.jp/job-feed/public-advertising/zx64umk

    The first DCON, chaired by Dr. Matsuo, was held.

    The DCON2020 (First National College of Technology Deep Learning Contest), of which Dr. Matsuo is the committee chair, held its final competition this August, with 11 teams competing.

    DCON is sponsored by Japan Deep Learning, a general incorporated association, and is a contest in which technical college students create works that combine the “technical skills of manufacturing” and “Deep Learning” that they have cultivated on a daily basis, and compete for the business potential created by their works in terms of corporate valuations.

    The DCON2019 teams from Nagaoka National College of Technology and Kagawa National College of Technology have contributed to the production of startups from technical colleges, including startups from two of last year’s DCON2019 participating teams, Nagaoka National College of Technology and Kagawa National College of Technology.

    The second DCON (DCON2021) was held last month in November and 30 teams were selected for the document review. The main competition is scheduled for next April 2021.

    These are the main news items of the Matsuo Institute in 2020!

    Please have a happy New Year.
    We look forward to working with you in the New Year.