1. Home
  2.  > Programs & Courses
  3.  > Graduate School of Science and Engineering : Information and Computer Science (Laboratories : Intelligent Systems Design Laboratory)

Graduate School of Science and Engineering
Information and Computer Science

Intelligent Systems Design Laboratory


Mitsunori MIKI
Telephone : +81-774-65-6930
Office : KC-121
Database of Researchers
Hiroto AIDA
[Assistant Professor]

Telephone : +81-774-65-6780
Office : KC-123
Database of Researchers

Research Contents

A PC cluster is a parallel computer connecting multiple commercially available PCs by a network. Compared to supercomputers with the same level of performance, the high cost-effectiveness of PC clusters can be given as its strength. In the Cluster Group, we are constructing PC clusters, evaluating their performance, maintaining and managing them, and developing clustering software. The Cluster Group is a research group pursing technologies that evolve daily. In the Intelligent Systems Design Laboratory, we are advancing cluster research by constructing Japan's largest PC cluster.
A grid refers to the systems and technologies to unify computational resources, human capital, and other resources that exist distributed in remote locations in order to use them as a single system. Through the development of grid middleware that can integrate multiple computers, applications, and services located distributed on a wide area network, the Grid Group is developing systems that can solve large-scale, complicated optimization problems such as structural optimal design. We are also participating in representative grid test beds such as ApGrid and OBIGrid and by providing large-scale PC clusters, we are contributing to grid research inside Japan and overseas.
Simulated Annealing
Simulated Annealing (SA) is an optimization method that simulates annealing in an attempt to obtain a superior crystal structure by gradually cooling materials melted at high temperatures. In the SA Group, we are improving SA with parallelization/decentralization, other optimization methods, and hybridization with evolutionary computation. SA is also applicable to actual optimization problems represented by LSI wiring design. In the SA Group, we are applying SA to actual optimization problems such as applying SA to the optimum design of Gain Flattening Filters (GFF)*.

A filter that has a function to smooth out variations in the amplification amount that differs according the light's wavelength
Genetic Algorithms
Genetic algorithms are optimization algorithms that simulate the process of biological evolution. By using the target problem's candidate solutions to resemble individual organisms and applying operators such as genetic cross over and mutation/natural selection to them, the candidate solution evolves and we can obtain the optimal solution. We are also investigating parallel models for genetic algorithms and conducting broad research on implementing genetic algorithms on PC clusters.
Interactive Genetic Algorithms
In the Interactive Genetic Algorithms Group, we are conducting research using Interactive Genetic Algorithms (IGA), one of the interactive evolutionary computing methods, as a technique for optimization based on human sensibility. We are proposing sign sound generation systems using IGA to create sign sounds used in household appliances and proposing Global Asynchronous Distributed Interactive Genetic Algorithms (GADIGA) as a technique to expand IGA into a massive participation model.
Protein Structure Prediction
The mapping of the human genome, worked on through international cooperation, was completed in April, 2003. Now clarifying the function of proteins is important. The function of a protein is closely connected to its structure and various results are expected such as discovering the formation mechanism of disease that occurs from a protein's mistaken folding (Alzheimer's disease, bovine spongiform encephalopathy) and the development of new drugs with the function of specific proteins. In our research group, we are predicting the structure of proteins with computers using systems that combine a technique called the molecular simulation method and optimization methods.
Multi-objective Genetic Algorithms
Multi-objective optimization problems are problems where the optimal solution is sought from multiple evaluation criteria that have trade-off relationships. In these problems, due to their characteristics, solutions exist as multiple solutions or a set of infinite solutions. In recent years there has been much research on multi-objective Genetic Algorithms (GA) that applies GA to multi-purpose optimization problems. In this group, we are proposing GA to obtain highly accurate solution sets widely distributed in a solution space, and we are conducting research such as the optimization of diesel engine fuel injection scheduling.
Web Communication
Our purpose in the Web Communication Group is to create a system to support the communication of teachers and students in the laboratory to energize research activities. The created system uses blogs and supports creating "connections" between blogs.
Intelligent Lighting Systems
Intelligent lighting systems are systems in which individual lights control the illuminance for respective locations by autonomous learning. Since they have no centralized control mechanism, the system has a high fault tolerance and achieves high reliability in large-scale buildings. The system can automatically judge the effective illumination and supply a suitable illuminance in appropriate locations just by users setting the target illuminance for the illuminance sensors, without requiring the illumination's or illuminance sensor's location information. These next generation illumination systems are attracting attention because they can realize energy savings by avoiding turning on unnecessary lights.
Swarm Intelligence
In recent years swarm intelligence has been gaining attention because although individual intelligences are simple, extremely sophisticated intelligences emerge when these gather and form groups. Swarm intelligence can be viewed in the societies of living organisms. In ants for example, even though individual ants behave simply, as a whole they behave intelligently to efficiently gather food. In our research we are developing algorithms to make this kind of swarm intelligence emerge and applying it to swarm robot behavior learning.


  • Intelligent systems
  • Intelligent home appliances
  • Optimization
  • Evolutionary computation
  • Parallel processing

  • Parallel computer
  • PC cluster
  • Grid
  • Web communication
  • Bio-informatics