Research Lab

Discrete Algorithms

The center of the recent ICT, such as search engines, block-chain, and so on, that supports infra-structures of recent society, consists of computer programs, and "algorithms" are principles how they works. Theory of algorithms is a new kind of mathematics that is required in computer science, and we study those theories for designing and analyzing correct and efficient algorithms and discrete structures for which such algorithms work. We also model practical problems that have discrete structures, and implement computer programs based on algorithms that we develop to solve such problems.


Social Informatics

We analyze various social problems and phenomena using game theory, stochastic models and simulations. Our research explores the value of information and its consequent impact on behavioral strategies. We are working on research with the aim of making support for deriving the best decisions among multi-agents in social systems and offer of mechanisms to improve social efficiency. Simulations are based on pseudo-models and often do not accurately reflect the real world. In inference based on data, it is possible to estimate only a few parameters. Applying data assimilation, the accuracy of the simulation can be improved, and it is possible to make higher-dimensional inferences. In addition to data assimilation, we are exploring the statistical aspects of machine learning methods.


Human Information Systems

While technologies for collecting and accumulating a wide variety of data have been built, the development of technologies for analyzing those data and utilizing them for decision-making is insufficient. Data Mining is a technique for finding useful knowledge from data. We aim to create a human-friendly information society through research and development of systems that support user decision-making by making full use of multivariate data analysis and soft computing technology, including recommendation systems based on historical data on the Internet.


Computational Intelligence

Real-world optimization problems often have multiple objectives to be optimized simultaneously. There are trade-offs between the objectives, and there is no single optimal solution. For such problems, we are developing evolutionary multiobjective optimization methods that can efficiently search for multiple Pareto optimal solutions and provide them to users. We apply the methods to knowledge acquisition by considering the trade-off between accuracy and explainability.


Emergent Software

After a long evolution, humans got intelligence and created lots of things. We try to make a computer system that understands and generates such people's creations! We use the latest technologies in image recognition and natural language processing based on artificial intelligence, machine learning, and evolutionary calculation. Our research topics are all exciting: manga, novels, picture books, photographs, starry sky, music, mahjong, and games!


Intelligent Signal Processing

Research topics are analysis of still images, self-position estimation and acquisition of environmental information from movies based on intelligent signal processing methods such as deep learning. Efficient model design that reduces calculation costs while maintaining accuracy are also interested.


Intelligent Media Processing

While the goal of Artificial Intelligence (AI) is to create machines that can replace humans, the goal of Intelligence Augmentation (IA) is to amplify human capabilities with the help of machines. I am challenging to realize intelligence augmentation through sensing human behavior and generating dynamic and intelligent media based on the results of sensing. In particular, I have recently been working on the analysis of human learning and the transfer of human experience.

→Go to Lab's Homepage


Emerging Communications Technologies

We are interested in molecular communication – a new information and communication technology paradigm. Unlike current electromagnetism-based communication, molecular communication uses chemical signals or molecules as the carrier of information. Molecular communication allows a group of bio-nanomachines to communicate and cooperate. Potential applications of molecular communication include creating an artificial information network inside our body.


Smart Platform

Our laboratory focuses on a creation of various services over programmable platforms and R&D on its underlying technologies. Toward smart society through ICT, it is important to develop an information platform which is capable to compose various functions freely and dynamically. We design a new platform which can control, deploy, and cooperate of computing resources, networking, and variety of data exchanged. Our aim is also its application to create practical services in the university.


Intelligent Networking

We are working on the realization of network mechanisms and controls that can provide ultra-high-speed, high-capacity, and low-latency information communication as well as smart and secure services as future networks. Our research is directly related to optical networks, wireless networks, network security, IoT and sensor networks, and edge computing, etc. 

→Go to Lab's Homepage


Advanced Computing Infrastructure

We are studying parallel programming for parallel computers that can perform multiple operations simultaneously. Most computers today, from smartphones to supercomputers, as well as PCs, are parallel computers. The number of operations that can be executed simultaneously tends to increase rapidly. In order to get higher performance out of them, parallel programming that explicitly specifies the simultaneous execution of multiple operations is necessary, rather than sequential programming that only specifies one operation at a time.


Probabilistic Information Processing

A time series is a series of data points indexed in time order. Time series arises not only in informatics and engineering but also in various fields. For example, a series of economic data like foreign exchange fluctuations is a time series. We are making researches on time series to extract important information and to predict future values of some phenomena. We are constructing models from time series, using machine learning. We are also developing signal processing technologies for future advanced information communications.


Control and Artificial Intelligence

Our research focuses on mathematical modeling, control design, learning and optimization of large complex systems composed of interacting agents with discrete, continuous, or hybrid dynamics. Collaborating with world-leading universities and scientists, our work is being widely applied in autonomous robots, self-driving vehicles, mobility networks, logistic automation, and smart cities.