Research areas that are co-created with society

At our university, under the unique academic field of Systems Information Science, the diverse expertise of our faculty members—including information systems, security, artificial intelligence, robotics, cognitive science, education, IoT, XR, design, biological sciences, complex systems science, space, and game theory—combines to create research areas that are truly characteristic of our university.
One of its distinguishing features is that faculty members from different research fields collaborate to form interdisciplinary research teams within the university, actively venturing into new research areas.

In its first 10 years, the university designated "Mobile IT," "Marine IT," and "Medical IT" as its three key research areas (the three MITs). Since then, the university's efforts have extended to collaborations with the local community and the world, as well as to creative exploration that integrates and deepens the knowledge of information technology and intelligence research that it has accumulated over the years.

Here, we introduce research examples from our university that use logical thinking to find answers to the numerous "unsolvable problems" that exist in the real world.

Information technology is changing society. This statement has become a tangible reality thanks to the rapid proliferation of mobile phones, smartphones, and wearable computers. Behind this lies the miniaturization and high integration of intelligent computing circuits, as well as the networking of vast amounts of data from society as a whole, known as big data.
We aim to build 21st-century infrastructure that every citizen can carry with them, by linking the evolution of information technology with new social services. The era of top-down infrastructure construction by the government is over; what is needed now is to listen to the voices of citizens—the users—and deliver customized services that are tailored to their needs. This initiative to revitalize local communities facing a super-aging society into "smart cities" made possible by information and network technologies is what our university calls Mobile IT, and we are conducting research and development on it as one of our priority areas.

As one of the central research areas,2012 2012 We launched a project to explore new forms of public transportation using mobile IT. We established the "Smart City Hakodate Lab" on campus and founded the "NPO Smart City Hakodate" as an organization to promote social practices in the Hakodate area.2013 2013 In the fall, using Hakodate City as a field for demonstration experiments, they successfully conducted operational trials of the world's first fully automated, fully demand-responsive public transportation system, "SAVS (Smart Access Vehicle System)."

This initiative to integrate the fisheries and marine sectors with IT is a highly unique research area, named Marine IT by our university, and is the only one of its kind in the world. We are working on developing and implementing technologies to accurately understand the ever-changing nature of the ocean, influenced by seasons, time of day, and weather, as well as the abundance of marine resources. Our university has designated these activities as Marine IT and positioned them as one of its priority areas.
For example, the ubiquitous buoy system, which can automatically and remotely observe water temperature and currents at each depth, has created a groundbreaking operating environment in which fishing and aquaculture businesses can check data in real time on their smartphones or tablets without having to go out to sea themselves.

Furthermore, by controlling catches through the sharing of vessel location information and fish school information among fishing vessels, overfishing can be prevented, enabling sustainable fishing for future generations based on appropriate resource management. Resource assessments, previously performed analogically using onboard blackboards, nautical charts, and radiotelephones, have been digitized and replaced with tablets and GPS (Satellite Positioning System), enabling more streamlined and convenient operations. Furthermore, fishermen can now access accumulated data to use as a reference for operations, analyze past data to understand resource conditions, and perform other analyses previously only available at fisheries research stations, enabling proactive and strategic initiatives. Collaborative efforts with local businesses are spreading throughout Japan, including Rumoi City and Fukushima Town in Hokkaido, and overseas to South Korea and Bali, Indonesia.

While the study of narratives is prevalent in the humanities, research that uses scientific methods such as data science to reveal how narrative patterns are constructed is extremely rare, making it one of the unique research areas at our university. If we can scientifically analyze high-quality works and reveal how the appeal of a story is constructed, it will not only be useful for creators but will also have applications in various fields such as AI and computer software.

Modern entertainment works such as manga, anime, and video games have complex narrative structures to keep readers and viewers engaged. However, research has revealed that even seemingly complex and difficult-to-understand narrative structures can actually be quantified as combinations of several basic short story patterns, and their characteristics can be analyzed using computers.

At our university,2012 2012 We are conducting research on the automatic generation of stories with a punchline by analyzing approximately 1000 short stories by Shinichi Hoshi. Furthermore,2023 2023 In the NEDO (New Energy and Industrial Technology Development Organization) project (TEZUKA2023) to have AI create a new Black Jack story, I was in charge of extracting (mathematically modeling) and automatically generating the narrative structure characteristic of Tezuka's works.

This project aimed to explore how close current AI can come to capturing the "interest" and "charm" of Osamu Tezuka's work, as well as to the extent to which AI can understand the creative process, whether it can become a supporter of human creativity, and to discover its limits and potential. In other words, it was an effort to explore not only the collaboration between AI technology and humans, but also the "possibility of AI supporting human creativity."

Furthermore, we are developing a creative support AI system that can automatically generate stories of various genres, building upon these achievements. In addition, we are using data science methods to implement various functions useful for story creation in this creative support AI system, such as converting short stories into novels, suggesting sequels to existing stories, and suggesting appealing characters.
Going forward, we plan to scientifically unravel the appeal of stories and explore the possibilities of AI supporting human creativity.

While artificial intelligence (AI) has made remarkable progress, much of it is based on simplified mathematical models of the brain's information processing principles. However, the actual brain is an extremely dynamic system that continuously updates itself in a constantly changing environment, integrating sensory information, predicting the future, and generating movement. The fundamental principles of such time-evolving intelligence are still not fully understood.

Our university's brain-inspired artificial intelligence research is a challenging endeavor based on mathematics, physics, and information science, aiming to elucidate the principles of this "dynamic intelligence" and reconstruct it as an artificial system. It's not simply about improving performance, but directly confronting the fundamental question of how the brain understands, predicts, and acts upon the world.

For example, we are building a mathematical model that integrates reservoir computing, which utilizes the dynamics of recursive neural circuits, with predictive coding theory, which explains the hierarchical structure of the cerebral cortex. By integrating multiple sensory information such as sound and vision, we have realized a model that can recognize speech with high accuracy even in noisy environments. This model demonstrates the ability to flexibly adapt to environmental changes by minimizing prediction errors while maintaining the temporal context as an internal state.

Furthermore, we are developing experimental research to implement this theoretically constructed computational principle in cultured neural networks using living nerve cells and various materials with nonlinear response characteristics. The "physical reservoir," which directly utilizes the dynamics of neural activity and the hysteresis effects of materials as computational resources, presents a computational paradigm different from conventional digital computing. In this paradigm, information is not simply processed as symbols, but represented as the time evolution of physical states.

These results lead to adaptability that allows learning even with limited data, the ability to handle time-dependent changes inherently, robustness to noise, and energy-efficient information processing.

Our university's brain-inspired artificial intelligence research is a comprehensive research area that spans theory, experiment, and implementation, simultaneously advancing our understanding of the brain and innovation in AI technology. By returning to the principles of intelligence and delving into its physical foundations, we aim to redefine the nature of next-generation artificial intelligence.

Recent intelligent robots possess high environmental adaptability due to their ability to mimic the physical functions of living organisms. However, because they are equipped with computers, they cannot operate sustainably in environments without an energy source such as a power supply.

Therefore, focusing on the fact that living organisms perform all information processing and movement solely through the chemical reactions of ingested organic matter via "metabolism," we have been developing a "chemically intelligent robot" equipped with the intelligence to move autonomously according to the situation using only chemical reactions.

By combining the Belousov-Zhabotinsky (BZ) reaction, which condenses the metabolism of living organisms, with a hydrogel (BZ gel) that swells and contracts accordingly, we created a chemical robot that performs peristaltic movements in any direction depending on the direction of reaction propagation, thereby realizing basic "sensory and motor" functions.

Furthermore, we are theoretically elucidating the mechanisms of these functions through numerical simulations using mathematical models, and promoting the implementation of artificial intelligence that realizes "judgment and learning" functions solely through chemical reactions. Specifically, we have demonstrated that the "BZ Reservoir," which considers the BZ reaction as a neural network similar to the brain's nervous system, possesses advanced learning capabilities. Currently, we have succeeded in recognizing and plotting the numbers being read aloud from spoken audio.
In this way, the computational infrastructure based on chemical reactions, which is necessary for realizing autonomous robots, is steadily being established.