The “DL Round-Reading Group” is a study group to catch up on the latest papers on DeepLearning.
Its memorable 256th meeting (2 to the power of 8) was held last month. (The photo is from an offline meeting held in 2019 or earlier.)
Today, we would like to introduce you to the activities of the DL Circular Reading Group!
What is a ︎◼︎DL Circular Reading Group?
The DL round reading group introduces and presents papers on Deep Learning every Friday morning at 10:00 a.m. The group started in 2014 and has been an activity for more than six years.
Two or three people will be in charge of rotating the papers to be introduced each time, and the participants themselves are free to choose the ones that interest them.
Participants include not only students, but also working professionals from major companies and venture companies. Some are involved in Deep Learning in their work, while others were involved in related research as students and wish to continue their studies even after entering the workforce.
Therefore, each session is full of questions and answers from a variety of perspectives, and the participants actively participate in the discussions.
Since the 2020 event was held online, it was open to course graduates from all over the country.
︎◼︎ ranking of presentation slides in terms of number of times shown on Twitter
Slides presented at the roundtable will be uploaded on SlideShere from time to time, and here are the presentations with the most views during 2020.
*Total from the “DL Hacks” Twitter account from January to December 21, 2020.
5thWhat do Models Learn from Question Answering Datasets?
Presenter: Kazutoshi Shinoda
No. 4 YOLOv4: Optimal Speed and Accuracy of Object Detection
Presenter:Koichiro Tamura
No. 3Set Transformer: A Framework for Attention-based Permutation-Invariant Neural Networks
Presenter:Mr.Takumi Okuma
No. 2Reformulation of Reinforcement Learning
Presenter:Mr.Yusuke Iwasawa
No. 1Summary of Recent Offline Reinforcement Learning -Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems-.
Presenter: Tatsuya Matsushima
︎◼︎256th DL Roundtable Participant’s Comments
Presentations were given by Masahiro Suzuki, Project Assistant Professor in Matsuo Laboratory, and Tatsuya Matsushima, a first-year doctoral student. About 70 people attended the event, including current students, graduates, and those who had not attended a DL reading session for a while.
Norihisa Kobayashi Present Square Co.
I have been a continuous participant since I attended the DL course (basic and applied) about a few years ago.
At the 256th presentation of papers and research, I was able to listen to hot topics such as world models and off-line reinforcement learning, and the reception provided an opportunity to share research topics and approaches of the participants. I have learned a lot and been stimulated by the daily round-reading sessions, but this time it was also very stimulating to discuss how Deep Learning is being used and the future direction of Deep Learning, not only within the framework of the papers. Thank you very much.”
Junpei Arima,Master 2, Graduate School of Science and Engineering, Meiji University
I have been a member of the “reading group” since the Matsuo Lab summer school held two years ago. In the commemorative 256th meeting, Mr. Suzuki and Mr. Matsushima gave a comprehensive presentation on world models and offline reinforcement learning, which have been active areas of research for the past few years, and I was able to deepen my understanding of these topics.
The reception after the presentations was a great opportunity to interact with a wide range of people from different backgrounds who were participating in the round-reading sessions, and we had a very productive time.”
◼︎ Facilitator: Mr. Yusuke Iwasawa, Project Lecturer
Mr. Iwasawa, a specially-appointed lecturer, facilitates each session in a fast-paced manner.
He is also available for consultation when you are not sure which paper to choose.
2012 Graduated from Sophia University, Graduate School of Science and Technology, Department of Informatics
2014 Completed the Information Science area of the Graduate School of Science and Engineering, Sophia University
2017 Completed Doctoral program in Technology Management Strategy, Graduate School of Engineering, The University of Tokyo (JSPS Research Fellowship DC1)
2017, Project Researcher, Matsuo Laboratory, The University of Tokyo
2018, Project Assistant Professor at the same laboratory, The University of Tokyo
2020, Project Lecturer, Do Laboratory, The University of Tokyo
[Research Interests].
Machine learning/deep learning applications to IoT (currently, deep learning models, especially considering privacy, fault tolerance, power consumption, etc.)
Representation learning for action recognition with wearable sensors (Ph.D.)
Automatic road surface accessibility information collection system using wheelchair sensing (M.S.)
︎◼︎ for those who would like to participate in the study group.
★About Regular Study Sessions
We periodically offer tours to those who have attended lectures and seminars held by Matsuo Lab.
Please come and visit us first!
Various links
◆Books translated by Matsuo Lab. are available for purchase here.
SlideShare is a web-based community where you can share your reading materials with others.
Twitter account is here.