With the vision of “creating intelligence,” Matsuo Laboratory promotes research on Deep Learning. In particular, we are advancing research on world models, robotics, large language models, and Brain x AI.

Artificial intelligence is a field with a history of about sixty years. Still, research to date has revealed very little about the mechanisms of intelligence and brain structure. Of course, there is a lot of partial knowledge about the brain, but there are still many mysteries, such as how it can recognize, learn, speak, or be conscious in the first place. I would like to unravel these mysteries and understand the mechanism of human intelligence.

As for deep learning, the Matsuo-Iwasawa laboratory is developing basic technologies such as deep generative models, deep reinforcement learning, and representation learning. In particular, we have recently been working on world models and their application to robot operation, viewing them as an important technology for engineering intelligence. Matsuo Lab operates TRAIL (Tokyo Robot And Intelligence Lab) as a research and development community aiming to realize intelligence in the real world.

Regarding large language model research, the Matsuo-Iwasawa lab has released “Weblab-10B” and is conducting a large language model course that utilizes the knowledge gained from our research and development activities.

We are also engaged in research to elucidate the information representation, mechanisms, algorithms, and architecture that support the brain's computational functions based on the rapidly accumulating knowledge of neuroscience.

Many of the papers by our laboratory members have been accepted by top-tier international conferences (ICLR, NeurIPS, ICML), and our students have received awards, including the Dean's Award of the Graduate School of Engineering. We provide an environment where researchers of various specialties gather to engage in friendly competition. It is also an environment where motivated students and researchers can play an active role by holding weekly readings of the latest papers, lectures, and discussions with leading foreign experts and external specialists.

The laboratory also holds weekly readings on deep learning and a study group on implementation called DL hacks. Since 2015, we have been providing Deep Learning Lectures, which approximately 10,000 students and business people take annually.

Graduates of the lectures are widely active in the academic and industrial fields of artificial intelligence.

For more information on trends in artificial intelligence, the book “Will Artificial Intelligence Surpass Humans?” published in March 2015 would be a good reference. It won the Okawa Publishing Award and the Special Jury Prize at the Business Book Awards. In addition, our laboratory members translated “Deep Learning” by Prof. Bengio et al. of the University of Montreal and “Reinforcement Learning” by Prof. Richard S. Sutton.

Matsuo-Iwasawa Laboratory actively promotes lectures to share basic research results through high-quality education, providing valuable learning to as many people as possible.

The courses cover various topics, from the basics of deep learning technology and data science to entrepreneurship education. We are also one of the first to offer courses on the latest technologies, such as Large Language Models, providing opportunities to deepen understanding through hands-on practice.

In particular, the Global Consumer Intelligence Endowed Chair (GCI Endowed Chair), offered since 2014 with the cooperation of the Ministry of Economy, Trade and Industry, and corporations. The lecture was attended by more than 10,000 students, who had the opportunity to acquire basic data science and machine learning skills. The course is open to students from all over Japan (graduate students, undergraduate students, technical college students, high school students, and junior high school students) and students from the University of Tokyo. As for working adults, employees of the Metaverse Engineering Department's corporate members and those on leave due to life events (nursing care, childcare, etc.) are also welcome.

We aim to become a laboratory that conducts solid basic research for the future while contributing to the industry at large.

Matsuo Institute is actively engaged in joint research with private companies to return the knowledge of academia to society.

By applying the latest research and technology to business, we aim to bring significant value to society while accelerating research and development by utilizing the abundant and practical data held by companies. As mentioned above, Matsuo Lab. does not specialize in a particular model to “create intelligence” but brings together a diverse group of specialists to conduct research activities. When returning these technological seeds to society, we naturally do not focus exclusively on natural language processing or image recognition but use technologies that meet the various needs and work to solve the problems companies are facing.

Matsuo-Iwasawa lab also focuses on incubation activities, intending to produce 100 university-launched startups per year that promote digital transformation (DX) in various industries. 

Matsuo-Iwasawa Laboratory has supported the launch of startups such as Gunosy and PKSHA Technology. I personally serve as the chairperson of the DCON committee for technical colleges, and recently, we have seen some outstanding young people, including technical college students and junior and senior high school students, aspiring to start their own businesses. The Matsuo laboratory will continue to work to bring the latest technology back to society by fostering startup companies and realizing a new ecosystem through industry-academia collaboration.