Faculty members who are responsible for teaching research subjects at the Graduate School of Information Science Engineering Master's and Doctoral programs are as follows.
For applicants who are planning to apply for the Graduate School of Information Science and Engineering, please check the major and course which your desired supervising professor belongs. The major and course which your desired supervising professor belongs will be your intended major and course when making application.
Master's Program Computer Science Course
Master's Program Human Information Course
Primary research topics of the Computer Software Laboratory are operating systems, database systems, distributed systems and computer networks from a point of view of system software development. Currently we focus on the research area of sensor network systems as software infrastructure of ubiquitous computing environments. A sensor network system is composed of a large number of sensor nodes with sensors and wireless communication devices, which realizes intelligent and sophisticated tasks with coordination of numbers of nodes. Our current research projects include: personal sensing systems, sensor database, event-driven sensor network systems, high accuracy indoor localization systems, clustering techniques for energy saving, embedded systems, and P2P microstorage systems. We approach these projects from the aspect of sensor network and its related technologies.
The Software Engineering Laboratory researches methods of eliciting software requirements from clients or users, of analyzing the elicited results, of creating specifications from the elicited results, and of managing the specifications of such requirements (Software requirements are the information used to specify the functions, performance, design constraints and so on required for the software to be developed.) Among other things, the Lab is working on the research projects below.
- Using scenarios to help elicit requirements
- To find problems in functions and user interfaces, the Lab develops scenario languages and uses scenarios to express the behavior of software and users. This research work includes changing scenario points of view, generating scenarios from normal scenarios, generating scenarios by replacing parts of existing scenarios, and ways of assisting scenario classification.
- Using ontology requirements to help elicit requirements
- To assist software analysts without domain-specific knowledge who need to elicit requirements, the Lab prepares domain-specific knowledge as a requirements ontology. The Lab also researches ways of compiling the elicited results into semi-formol Japanese requirements specifications.
The Intelligent Communication Laboratory adopts a distributed model of computation that rests on a network of intelligent cooperating agents deployed in uncertain environments with a common goal of assisting and safeguarding humans.
Robots are controlled
by autonomous intelligent agents,
while a robot simulator is
used to test the agents
and inspect the robots' behavior.
- [Ambient Services]
- Using accelerometer and GPS information of smartphones, everyday patterns are estimated to evaluate and suggest individualized services. Based on the location and state of the user, the most suited services are selected. These services are implemented in control of household appliance and robots by using multi-agent systems.
- [Ad Hoc Social Networks]
- This type of network provides a local and temporary social network for data and information exchange or discussions. The communication is set up by only using smartphones without relying on Wi-Fi access points, etc. Smartphones that have no immediate connection with each other can use intermediate smartphones as bridges to transfer data and communicate.
- [Epidemiological Modeling and Visualization of Knowledge Propagation]
- To understand how information and knowledge is diffusing in societies, medical and biological methods to track diseases (viruses, bacteria, etc.) are used to model their behavior. Knowledge propagation has been simulated in scientific publications and classified by their cultural characteristics.
The Computer System Laboratory researches and develops both hardware and software designs to boost the performance of the computer systems that realize infrastructure for information society.
- Parallel processing is a key issue for increasing computer system performance. Superscalar processor executes multiple instructions simultaneously. We aim to reduce instruction-level dependency by speculative execution such as branch prediction and load value prediction.
Multi-core processor executes multiple processes simultaneously. The interconnection network is a key issue for decreasing communication overhead between cores. We aim to develop low cost and low latency routers for on-chip network.
- Data mining is an important technology to extract hidden characteristics from a huge amount of data. We aim to develop efficient algorithms for handling a huge amount of data, and applications for practical use. We are conducting research and development such as real time burst detection algorithm, efficient recommendation algorithm for EC site, filtering harmful web pages, net patrol systems, and so on.
We are researching how to improve networks by integrating wireless technologies into the Internet. Our research interests include mobile ad-hoc networks consist of mobile devices connected by wireless link, vehicular ad-hoc networks (VANET), multicast communications for multimedia distribution and network coding theory. Through these researches, we would like to develop new generation network technologies to achieve a secure and reliable IT-based environment.
Vehicle Visualization System on the iOS Devices using VANETs
The photo shows a vehicle visualization system which can display information about neighboring vehicles on in-car tablets. Tablets in each vehicles communicate with each other using wifi ad-hoc communication without infrastructure support, e.g. cellular network support. This system can support drivers to recognize invisible vehicles near for crash avoidance.
We are developing technology to efficiently search multimedia information such as photos, music, video, and actions, etc. We are forcefully promoting research into similar search technology for multimedia information and its applications to music databases. In addition, we are developing technology to do similarity search of photo, video and human moving trajectories, etc. Further, we are also involved in improving web information search technology. We are researching and developing data engineering (databases) and network services that efficiently manage and analyze vast amounts of information with the aim of achieving efficient information access in this era of information explosion. Finally, new paradigm of recommendation systems are studied: Context- aware Recommender Systems, Serendipity-oriented recommendation systems which recommend unexpected items, Recommendation systems based on user-generated contents in real world, and so on.
Focusing on the system programs that form the core of computer systems, the High Performance Computing Software System Laboratory is working on various research approaches that aim to achieve high performance computing with high reliability and high availability. The lab’s system software research and development is always orientated toward what is feasible and practical. Our work focuses on the following areas and approaches to research.
[High Performance Computing and High Availability Systems by using Parallel and Distributed Processing]
1. We are researching and developing ways to increase the execution speed and utility of advanced parallel processing/ distributed processing methods such as cloud computing and data centers through such things as load balancing mechanisms for multi-core and many-core environments, reliable server construction technologies, and Infrastructure as Code. The technologies are then applied to developments in the emerging field of GPGPU, numerical simulation of biological functions, big data analysis techniques, etc.
[Safe and Secure Systems]
2. We are researching concrete system software approaches to create safe and secure systems highly resistant to ‘Impersonation Attacks’ and ‘Process Hijacking’ (two of the major insidious software cracking techniques). In addition, the lab is involved in researching password vulnerability checks and reliable authentication technology.
[Software Engineering Approach]
3. Our research uses a software engineering approach to create highly reliable software, through both theoretical and applied research in areas such as program analysis, model checking, Depart concurrent computation models and type theory.
The Web is fast becoming a collaborative platform for the creation of new information and knowledge. In our lab, we are involved in achieving Web collaboration to support people's activities and communications on the network. In particular, we are developing specific applications by applying the Semantic Web technology and multi-agent technology. For example, we developed Yubisashi-chat ("Finger-Pointing Chat") to support communication with people with a cognitive handicap such as aphasia or dementia.
To support finding difficulty of people with aphasia, we also developed a word retrieval assistant prototype as a Web application, which controls the order of multiple choice type questions to ask and guesses the word an aphasia sufferer wants to say based on teh answers given.
Additionally, towards making use of knowledge of many people (collective intelligence), we are conducting research on gamified crowdsourcing for knowledge base acquisition and refinement.
Screenshot of “Yubisashi-chat (Finger-Pointing
Chat)” software prototype
Screenshot of "Word Retrieval Assistant"
"Mobile computing" is a key technology for creating an enhanced information space integrating the real world and the virtual world. We are conducting research by using mobile devices which are equipped with various sensors to improve our daily lives in the near future. The functions of mobile devices are further refined every year, and which is enabling us to develop common core systems using mobile devices that cleverly, casually, and gently support various activities in our daily lives.
We are concentrating our energies into our research to develop a distributed framework for mobile mixed reality (MR) systems that enable users to experience the real world enhanced with electronic information provided through mobile devices. We have developed mobile MR applications based on the developed framework: a campus guide, a wiring and facilities inspection system, and so on.
Experiencing MR campus guide
using mobile devices
In the future, we will expand the range of research into automobiles and UAV (Unmanned Aerial Vehicle)to think extensively about "mobile computing".
Advances in sensor technology and computer technology make it possible to gather many kinds of data from subjects going about their routine activities. These technologies make it possible to record the types of school texts subjects have read, the websites they have searched for, and even the types of objects they have come in contact with that day. These records are known as lifelogs. Analyzing them lets researchers infer information on subject likes and dislikes. The Data Engineering Laboratory’s research and development work is designed to use these inferences to create computer systems that provide users with the services they want, when they want them, before even being asked. If successful, such systems will be able to provide services perfectly in tune with user needs. The Lab’s goal is to create an environment that will let even first-time computer users master the use of such services. Applications could span a wide range of fields, including assistance for the elderly, task management, education, disaster readiness, and inferring user interests.
Today’s computing environments have not yet reached their goal of supporting collaborative human activities. The Distributed and Collaborative Systems Laboratory is working on distributed computing environments to support collaborative human activities in learning places, offices and urban areas. Examples of research achievements are as follows.
- “Dynamic Group Collaboration Environment” which exploits tablet terminals and its application to collaborative web search and a puzzle game for kids.
- Programming learning systems which enable kids to share programming works or workshop experiences on SNS. This research is jointly being conducted with an NPO.
- Detection of people’s movement using Bluetooth or Wifi signals emitted from smartphones and its application to information sharing.
"Cognitive engineering" is adopted to design human-centered systems based on models of human psychological and behavioral characteristics. This methodology is used to four major fields as follows:
1. Disaster Mitigation (Resident and tourist evacuation support, business continuity management, disaster information collection, and disaster mitigation education and training, etc.)
2. Intelligent Transportation Systems (driver support information system, tourist navigation system, support people who have no sense of direction, Sunday drivers support when they restart driving, and winter road management, etc.)
3. Sensitivity Engineering (fashion coordinate support through conversation with clothes, control of personal tempo to support individual cognition and action, and coaching support of street dance, etc.)
4. Fond memory engineering (fond memory recall support, communication support and knowledge transfer support by using fond memory, etc.)
Students learn methodology of cognitive engineering through experience of the whole process of needs mining, specification, designing, implementation, and evaluation of systems.
Mutual support system among
passengers trapped inside a train
DJ expertise acquisition support system
Research and development for fundamental technologies on providing services that integrate the real world with information spaces amidst the numerous computers and sensors and ubiquity of networks environment.
- Ubiquitous Computing, IoT
- Tablets, smart phones, wearables and hearables have been equipped with full of sensors and continuous connection to the internet. We developed smart phone based persistent sensing system along with the cloud hosts mining life-log data. Human activity sensing, recognition and near-future prediction research are investigated in order to realize TPO (Time/Place/Occasion) -based applications development.
- Real-World-oriented Systems
- Such technologies as organizing vast world of information contents concerning with real-world things and events and characterizing users preferences in accordance with accesses to those information are investigated as well as development of interactive client systems in a real-world-oriented way; in the underground city at the Osaka Station area, we have developed a panoramic viewer without pedestrians, barrier-free navigation, energy-harvesting indoor positioning system, Wi-Fi packet based human flow sensing system and ICT-assisted evacuation system for disasters.
- Embedded systems
- An embedded system is a computer system which is embedded to devices such as automobiles, industrial machinery, and electronics. The embedded system is one of the key technologies for ubiquitous computing and society 5.0. We are researching and developing real-time OS for embedded systems and distributed system for GPGPU computing, and participating OSS project(Autoware) for ADAS(Advanced Driving Assistant System) and its dynamic map generation and distribution systems.
- Global Information Networks
- On global information networks, all entities such as business sites, user terminals, mobile computers and sensors, can be connected to interact with each other without a centralized mechanism. Our aim is to establish fundamental technologies for autonomous distributed cooperative algorithms and to implement network applications and systems.
- Network Applications
- Since home appliances are expected to operate independently, we believe we should introduce distributed mechanism to home appliance networks.
Our network protocols combine multiple appliances cooperatively and control their use to support communications among family members.
- Overlay Networks
- An overlay network is a computer network built on top of another network. Our Japanese patent No. 433200 employs overlay network technology to the mechannism for WiFi Hotzone that covers a city with a wireless local area network.
Distributed home appliance network
- Mobile Networks and Internet Protocol
- A virtual single cell is our solution that supports high-speed roaming and mobility by unifying neighboring micro-cells. We can realize broadband mobile networks by combining this idea with Internet Protocol over WiFi Hotzone.
The goal of our research is to realize cyber world in which intelligent agents supports humans’ everyday life, by bridging the real world and the virtual world.
- Ubiquitous User Behavior Support
- Agents get the behavioral context of users using sensors, and provide appropriate services suitable to user's location and time.
- Recommendation System
- New paradigm of recommendation systems are studied; Context-aware recommendation systems, Serendipity oriented recommendation system which recommend unexpected items, Recommendation system based on user-generated contents in real world, and so on.
- Information Therapy
- Research on supporting communication and behavior of people with intellectual disabilities using information technology are studied. We have developed a system to present topic words on the screen which trigger the conversation with aphasiac user.
Our research group, headed by Prof. T. Maeda, carries out theoretical analysis as well as experimental investigations in the field of wireless intelligent communications for the efficient use of the frequency spectrum, a scarce resource, in order to realize the emergence of a universal society with access to ubiquitous computing. Professor Maeda's research activity covers a wide range of wireless communications.
The research topics include microwave antennas, body mount antennas, Software Defined Radio, UWB systems, Adaptive MIMO systems, microwave imaging techniques, and RF measurement techniques, including the human body’s effect on antenna systems. Since 2002, our research group has also been heavily involved in research into Ultra Wide Band wireless systems and extremely high-speed data exchange networks for a future society of ubiquitous computing.
In addition, our lab’s unique human equivalent phantom opens a new era in scale-model measurements, which enable scaling-up and down depending upon the practical requirements during the experimental investigations. These capabilities will allow our research group to develop new results, that few other institutions are capable of developing body area wireless networks as well as antennas for the systems.
Wireless communication system
test-bed for MIMO and SDR research
Self-adaptive power divider for reducing
the effects of the human body
Our research group are exploring principles and methods that can make software construction and evolution easier and faster. The group also track research work on design and implementation of next-generation software development environments embracing those principles and methods.
Latest research topics include: (1) Tool platforms that facilitate the construction of software-development support tools,(2) Mechanisms that can keep track of the evolution of a software application by analyzing its development history and visualizing its structure and behavior,(3) Automated tools of refactoring that improves internal structure of existing software without changing its external behavior to make the software easier to understand or cheaper to modify, and (4) Automatic evolution of software, such as an extension of functionality and bug repair based on development histories.
We research on system software, mainly operating systems (OS) and virtual machine monitors (VMM). System software is software that controls the core of computer systems, and so its power is paramount. Even if applications try to access to files, or even if they try to send data to other computers through network, OS can deny the accesses. Using this mechanism, we are developing a new OS that prevents personal information leaks even though applications have a security hole. Do you know applications which steal private information, such as address book on a smart phone? Only OS can stop it.
There are so many software called malware such as computer viruses, Trojan horses, etc. They are spreading through the world rapidly. And new types of malware appear every day. In 2016, over 357 million malwares have been found by certain institute. It is very important to analyze their nature to protect your computer. But we do not have enough time to analyze them because a new malware appears every 9 seconds. To solve this problem, we are developing an analyzer for malwares with virtual machine technology.
Let’s challenge to solve problems from view point of operating systems!
The New Generation Computing Laboratory works on methods of designing next-generation computers to meet the demands of contrasts with most computers to date, which tend to consider only the need to meet conventional demand for performance.
The Lab is also researching theoretical analysis of operating principles and ways of designing radically different computing methods from today's methods-technologies such as quantum computers and biocomputers. Related to the above-mentioned researches, the Lab is also researching theoretical aspects of algorithms and data structures, and parallel high-performance computing with GPUs, etc. Some of the Lab's current research areas are:
- Highly dependable computers
- -Research on ways of designing computers that can avoid failures
- Increasing LSI miniaturization has made the problem of hardware failures during manufacture and operation. This research is addressing ways of dealing with such hardware failures.
- Quantum computers
- -Creating methods and tools for efficient design of quantum circuits
- By controlling microscopic physical states such as electron spin, we can perform a type of parallel computation known as "quantum parallel computation." This approach results in computers that are significantly faster than current supercomputers for solving particular types of problems. This Lab is researching theoretical analysis of quantum computation as well as developing design methods and tools for quantum circuits, which are components of quantum computers.
Quantum superposition and a quantum circuit.
We are conducting research to solve various problems by integrating various IT and related technologies mainly robots to take action, computer vision to see events, artificial intelligence to judge the situation and instruct actions doing.
For example, in Intelligent Space research, data obtained from sensors distributed in space are processed to retrieve useful information then the space will provide comprehensive support; physical support by robot and informative support by image anad voice for people in space.
In addition to this, a wide range of researches in various fields such as a wearable robot arm that supports human work, an eldery care assistance robot, a on water environment monitoring system, and a medical behavior recognition technology are performed widely in this laboratory.
A nursing care support robot that
quantitatively measures nursing
care movement and teaches the correct
nursing care movement
Assistance Oriented Arm(Wearable Robot Arm)
Soft Intelligence Laboratory develops human symbiotic systems, evaluation and modeling of human sensibilities and emotions, decision support systems, tools for expert knowledge and skills acquisition, and novel methods of human-computer interaction using soft computing technologies such as Fuzzy Logic Theory, Neural Networks, Evolutionary Computing and Affective Engineering. The systems are developed to be "Affective-rich and Human-like" and to improve the standard of living rather than engineering. Examples of our research are as follows:
1. Human symbiotic systems aim for the realization of superior interpersonal affinities by the analysis of interaction anad the improvement of affinity in bidirectional interactive communication between humans and robots, humans and computers, and so on.
Specifically, our symbiotic studies are on (1) interactive emotional communication between humans and robots, (2) relaxation sounds generation systems that are based on multiple biological signals, such as brain waves or heartbeat data, and (3) intelligent instruction support systems for omni-directional wheelchairs with user intent recognition based on eye tracking.
2. In Affective Engineering, stimuli, for example movies or bento (Japanese lunch boxes), are evaluated based on adjective indices such as 'fun' and 'delicious-looking' rather than box office statistics or nutritional values. Next, models of human sensibility and emotional response are developed using the data collected and the soft computing technologies. The models developed may be used to support the production of a more fun movie or making a more delicious-looking bento.
Omni-directional wheelchair recognized the
human intention based on the eye tracking
The brain contains cells called neurons which number in the tens of billions, and which are thought to enable various functions through the trading of electrical signals called action potential, but the mechanism itself is still not well understood. We are researching this brain functions mechanism using mathematical methods such as theoretical analysis and computer simulations.
By modeling the electrophysiological characteristics of neurons, the aim is to clarify the informational processing mechanisms of the neural circuits comprising groups of neurons, and the mechanisms of information transmission at the level of the individual neurons. If this explication of the brain's information processing mechanisms advances, we hope to be able to apply it to developing methods of treating brain injuries and in interfaces that enable the brain to be accessed directly.
Our main study subject is “Mixed Reality,” which merges real and virtual worlds seamlessly in real time. We seek to improve the accuracy of geometric and photometric registration of real and virtual spaces. We are also expanding our research topics to “Auditory and Visual Mixed Reality” and “Diminished Reality.” Another research topic is “Human-Computer Interaction.” We look to develop post-WIMP (Window, Icon, Menu, Pointing-device) interfaces using “ToolDevices,” which expand on the intuitiveness and user's experience with real world tools, and “Wide-view digital working spaces,” using large screens and tabletop computers.
Concept of MR-PreViz
shooting in filmmaking
E-coloring using a brush painting device
We investigate the characteristics of human vision systems, and apply the acquired knowledge to color science and visual environment engineering. Understanding of human characteristics is important for evaluating the functionality and comfortability of industrial products or visual environments. Light that is input to the visual system is “physical”, whereas sight, which is the final output, is “psychological”, so we use psychophysical methods to examine the functions and relationships that link the two. So far, we have contributed to the creation of many products, including UDcolor®, software for the colorblind, CRS®, a lighting system for the elderly, and Feu®, a brightness index for visual environment. In addition, we are pursuing a wide variety of research topics such as motion picture sickness, visual acuity defined by cortex, assessment method of cataract progress, display color management system that does not require colorimeter, and evaluation of display image quality in various visual vironments.
- Gestural and Perceptual Interface
- Our vision-based intelligent interfaces including real-time 3-D hand detection and gesture estimation enable us to operate devices more intuitively.
- Indoor Surveillance System
- This system visually captures human activities in indoor scenes and recognizes them with computer vision. It can automatically detect objects a person places on or removes from a table. A user can ask "Who left this?" or "Who took the book from here?" through verbal and gestural interaction.
- Secretary Robot / Service Robot
- Secretary robot has ability to move around autonomously, identify human faces and engage in verbal conversations. It behaves as a physical agent mediating between a user and the indoor surveillance system. We develop one more mobile robot with manipulator which can open/close the refrigerator and pick up drinks. User can interactively operate the robot by verbal and gestural interfaces.
Indoor surveillance system
In our laboratory, we are involved in various research subjects for developing algorithms and systems to achieve 3D vision for computers so that it can be used as “eyes” for robots to operate in real environments. Some technologies developed in this laboratory have been used in products of 3D Media Co., Ltd. for factory automation. With an industrial robot, a mobile robot, a Nao humanoid robot, and various 3D vision sensors, our students enjoy research using these "expensive toys". With vision, an industrial robot can pick up, arrange and assemble industrial parts into products.
With vision, a mobile robot can walk around in our building. Other examples of the systems we have developed so far are a 16-camera 3D system for realtime tracking of field sports, mixed reality with a see-through stereo head-mounted display, SLAM with a stereo camera head, 3D scanning and integration system using Kinect, etc.
- Artificial Intelligence
- We focus on applications of artificial intelligence and other intelligent techniques to well-being games, video game live streaming, and digital humanities. If needed, basic research in relevant areas is also conducted. In the first theme, our aim is to develop intelligent game mechanisms, using, for example, Monte-Carlo tree search, for promoting physical, mental, and social well-being. For video game live streaming, intelligent techniques -- deep learning, etc. -- are being researched for automatically generating gameplays suitable for each spectator. In the third theme, we aim at promoting cultural content through exploiting state-of-the-art artificial intelligence methods such as recommender systems.
- Evolutionary Computation
- Here we study evolutionary computation (EC) that is the optimization technique inspired by the nature evolution mechanism. We explore EC algorithms to automatically generate game stages depending on players’ skill. We also develop EC applications to automatically adapt parameters of non-player characters.
We are developing various state-of-the-art methods related to computer visualization, virtual reality, and three-dimensional (3D) computer graphics. In addition, we are advancing several research projects to utilize our methods for researches of natural science, medicine, and digital archiving of 3D cultural heritages.
As for the field of natural science, by using output data of supercomputer simulation, we are visualizing the large-scale tsunami that occurred during the Great East Japan Earthquake in 2011. We are also conducting visual analysis of the tsunami that with a high possibility of occurrence in the future Nankai Trough Earthquake.
As for the medical field, we are realizing ultra-high-quality 3D see-through imaging of human internal organs based on large-scale medical data that are acquired by medical measurement devices such as CT scanners.
As for the field of digital archiving of 3D cultural heritages, we are digitally archiving festival floats and their background scenes of the Gion Festival in Kyoto, and we are reproducing them in the computer. We are also realizing high-quality visualization of big point-cloud data obtained through 3D scanning of the World-Heritage Buddhism temple ruins of Indonesia.
Human internal organs
Hachiman-yama float in the Gion
Festival, Kyoto Thanks: Hachiman-yama
Virtual Reality has been selected as the one of "Grand Challenges of the 21st Century" in the field of information technology, and is promoting the visual, aural, and haptic "pursuit of reality" to fill in the gap with the real world.
Whereupon, in our lab, we are aiming to generate virtual environments that enables various sensations and tactile information to be seen and felt using "cyber grasp", such as softness and holding something in the hand for deformable bodies such as human organs and woven cloth based on virtual reality, Ultra-realistic communication and Computer Vision technologies.
In addition, we are also conducting research into achieving "Multi-points haptic collaboration" through ultra-realistic haptic communications that share such “Tangible” virtual environments via networks.
Currently, two national leading projects are in progress in our lab; "Multi-points hands-on training system for Minimally Invasive Surgery" supported by MIC(Ministry of Internal Affairs and Communications-JAPAN) and "Mixed reality- based Digital Museum", supported by MEXT(Ministry of Education, Culture, Sports, Science and Technology-JAPAN) based on this research success.
Human intelligence is never designed directly, but it is grown. Babies cannot walk and speak languageds when they are born, however, over time they acquire these behaviors by interacting with others and environments and by imitating others' speech and movement. To understand human intelligence and dynamics of society, it is necessary to reveal properties of emergent intelligence.
To understand human intelligence we are employing constructive approach. We adapt several techniques including machine learning, robotics, and agent simulation to imitate the human developmental adaptation process. In addition, we are also focusing on the application of intelligent information technology to society and the communication of the various natural languages.
- Medical Image Processing and Computational Anatomy
- In recent years, computational anatomy has become a hot topic in medical image analysis research fields. We are working on the construction of a digital atlas of human anatomy such as the liver. The digital atlas can capture an organ's variability, such as its position, shape, and voxel intensity (texture) from a training set, and then be used for computer assisted diagnostics.
- Automatic Image Annotation and Image Indexing
- With the development of digital imaging technology, more and more information is nowadays conveyed in the form of digital images or video clips. We are developing an automatic image annotation method/system in which computer vision and pattern recognition techniques are used to understand the semantic meaning of an image. Metadata, such as captions and keywords can be automatically annotated to the image for image indexing.
- Facial Image Processing and Beautification
- Quantitative characterization of facial appearance is an important issue in many fields and applications including cosmetic foundation design. We have constructed a multi-angle view, illumination, and cosmetic facial image database (Ma-VIC) for appearance studies. Several morphing techniques have also been developed for beautification studies.
Surgical navigation system for treatment of liver tumors
The main research topic is computer simulation of physical phenomena like deformation of solids, flow of fluids and propagation of electromagnetic waves. Our approach is to combine the simulation with the techniques of computer graphics (CG) in order for accurate expression of three-demensional geometry, intuitive operation and high visibility.
We developed original techniques for shape representation of three-dimensional surfaces that contribute to accurate and efficient computation of physical phenomena. Some of our simulation techniques allow users interactive manipulation during the simulation. In fluid simulation, users can modify obstacles in three-demensional domain and, in electromagnetic simulation, users can change shape and material parameters of objects located in the space.
In addition, our shape representation is designed for effective parallel computation and efficient visualization.
Three-dimensional shape modeling
We are researching the analysis, understanding, reproduction, and synthesis of acoustic sound environments using media technology for building a more pleasant living environment. In particular, we research on daily basis topics in the area of "Acoustic Sound", such as "Audio Spots" based on spotlights of acoustic sound, "Acoustic Sound Sensors" based on abnormal sound detection techniques, "Hands-free Voice Interfaces" based on the distant-talking speech capturing techniques, "Optical Laser Microphone" based on the sound reproduction techniques by scanning from vibration objects with laser, "3D Acoustic Sound Field Reproduction" based on highly realistic acoustic sounds field reproduction techniques, and "Acoustic Planetarium" based on the design techniques of sound images with parametric loudspeakers.
Furthermore, we also try to actively contribute to R&D for new acoustic systems based on the auditory scene analysis of human beings.
1. Data recognition and generation: Deep neural network is used for data-driven learning to classify 2D or 3D data. Our research targets include image style transfer, 3D building structure design and 4D prototype analysis of a biological system.
2. Optimization: Several approaches from mathematical programming to meta-heuristics are used for the optimization. Our research targets extends from VSLI floor planning by simulated annealing, invetory control by demand prediction for optimal logitics, to power management in a decentralized network of renewable energy by stochastic programming.
3. Image analysis: Dynamic 3D facial expression images are important in many computer vision and graphic applications. We reconstruct the dynamic 3D facial images from only one facial image by machine learning.
4. Bioinformatics: Machine learning is also used in bioinformatics. One of the present targets is phenotype analysis using 4D dynamic data on cell division in early embryo of C.elegans as a joint research with a biologist group.
The laboratory researches interaction design methods among humans and information, in order to support for discovering new knowledge in creative activity.
- [Text Data Mining]
- Text data mining methods are used for discovering frequent appearing information and less-frequent but valuable information from the huge amount of data. The methods can predict new commercial items and new research themes that will become popular.
- [Information Visualization]
- Information visualization methods are used for understanding easily what data is. The methods transform numerical data and text data into visualized images and graphs. The below figure visualizes a text by using metaphor of light and shadow in a photo. Sentences related to a theme of a text are lightened. Not-related sentences are shadowed. The distribution of light and shadow shows a readability of a text.
A visualization system of a text by using
metaphor of light and shadow in a photo.
- [Conversation Analysis]
- Conversation analysis methods are used for detecting activated parts of a conversation and extracting important utterances. The methods support for making meeting minutes.
In the physiological engineering lab, we research and develop physiological engineering in relation to human behavior. We research lifestyle design and successful aging focusing on biological signals such as human physiological function, psychological factor, and behavioral activity.
In lifestyle design research, our aim is to propose new lifestyle habits by developing technologies in relation to sleep and alertness that are affected by physical stimulus from environment. The main topics of sleep research are sleep state detection, sleep-wake induction, and bedroom amenity. The main topics of alertness research are activation of brain functions, power of concentration, cognitive functions, and biological rhythm.
Physiological functions training test
with a horseback-riding simulator
In successful aging research, our aim is to improve and maintain flexible physical control capabilities by developing technologies in relation to training and relaxation of mind and body. The main topics of training research are muscular strength,
postural sway, fitness, and biomechanism.
The main topics of relaxation research are autonomic nervous functions, entrainment, and napping.
Brain functions activation test
Professor Hachimura has for a long time been engaged in research activities in the field of computer applications to the humanities, based on image processing and CG. Currently, he is conducting joint research with humanities researchers mostly from the School of Letters at the Art Research Center, Kinugasa campus. He is one of the leaders in "Digital Humanities Center for Japanese Arts and Cultures", which fosters links between the humanities and information technology research.
Because of this, the research topics in the lab are characterized by in-depth research into humanities-related fields. The current main research topics include image data analysis of historical paintings and documents such as Ukiyoe paintings and classic books, and digital archiving and analysis of intangible cultural assets such as dance by using motion capture.
Motion capture with gaze measurement
of Japanese dance
Float processing in the
Virtual Gion Festival
We are conducting researches on human-computer interaction, entertainment computing, and affective computing, based on Natural Language Processing (NLP) which aims to handle human language on a computer. We also interested in affection and emotion processing relate to NLP.
- [Human-computer interaction]
- Question Answering (QA) is one of the interaction support technology which extracts answer strings from a large amount of document data on the Internet. We are expanding this QA technology to handle metaphorical expression and system interaction in an ambiguous situation. We are also doing research on multimodal interaction such as gesture to realize more human-like communication in robots.
- [Entertainment computing]
- We are conducting tourism information processing for sightseeing assistance by extracting appropriate information from a variety of Blog sources and reconstructing the extracted information for personal travel assistance. We are also conducting cartoon information processing. We are now analyzing cartoon specific expressions and applying this to other domains.
- [Affective computing]
- We are conducting research to understand metaphorical expressions of taste information. We try to understand semantics of taste expression to compute taste expression similarity and generate appropriate taste expressions.
We are doing research mainly on information access technologies to facilitate the organization and use of a vast amount of text media information. In recent years, an increasing amount of information has become available online, but it is becoming difficult to find the desired information from it.
In order to support users in accessing information that matches his/her needs, we research techniques for information retrieval, which is the basis of information access technologies, as well as data mining techniques to discover useful knowledge hidden in a vast amount of information, information recommendation techniques that recommends useful information according to the user's interests, search and mining techniques for valuable historical materials that have been increasingly digitized, and multilingual information retrieval which enables access to information written in various languagesin the world.
Transition of personal relationships estimated
by analyzing historical documents
Interpersonal relationship diagram obtained
by analyzing a historical document
The Biorobotics Laboratory has developed wearable robots and soft mechanisms actuated by air pressure. In addition to these studies, the laboratory has recently started to study cognitive psychology dealing with human emotion, perception and motor control. Human emotions and decisions are unconsciously affected by the environment. Human perceptions also vary according to the environment. Understanding the relationship between the environment and the human mind contributes to the development of humanfriendly systems. Current research interests include preference formation, proprioception, haptic sensation and pneumatically controlled machine elements that change the rigidity.
Force display system
Odor preference judgments
Force display by the use of
constrictive pressure on forearms
Variable stiffness sheet
We are developing fundamental techniques and applications for effective use, transmission, and generation of audio information such as speech and music.
(1) Speech conveys paralinguistic information, which includes emotion, intension, attitude, and so on, as well as linguistic information. We extract the paralinguistic informaton from the user's utterance to realize a spoken dialog system which can know "what a user really wants".
(2) Speech is uttered with different prosodic expressions according to the mental state of a speaker even if the linguistic contents are the same. We aim to realize a speech synthesis system which can generate expressive speech with various voice qualities.
(3) Audio information plays important roles in entertainment applications. The research themes include synthesis and retrieval of a sound signal appropriate to special effects, and modeling the relationship between musical performance and body movement.
(4) We aim to recognize audio signals emanated in environments of daily life to understand human activities using audio information.
(5) We develop a method of information transmission for the close range using audio signals by embedding information in high frequency bands of the signals.
Human Robotics Laboratory conducts research on advanced intelligent mechanical systems that have physical interactions with humans such as rehabilitation robotics and automobiles. For realizing comfortable and usable mechanical systems, our research areas span 1) human modeling: understanding human cognitive and motion function, 2) robot control: control theory for mechanical systems, 3) human machine interface design.
- [Human Modeling]
- ・Computational model of self-motion perception of humans
・Computational mode lof motion sickness
・Understanding motion information of multi-task systems:extraction and synthesis of inertial motion
- [Robot Control]
- ・Control of transfemoral prosthesis to achieve comfortable motion based on inertial motion generation method
・Control of Human Machine Systems
・Finite time stabilization of robot manipulators
・Shared control of underwater robot
- [Human Machine System]
- ・Control methodology to smoothly connect automatic operation and manual operation
・Realization of driver assistance method of automobiles to simultaneously achieve workload reduction and skill improvement
ICT has become crucial in our society. The social impact of failures that occur in critical information systems is quite serious. In particular, cyber crime has become a major obstacle for threatening the safety and security of our society We are doing the research and development of technologies that will contribute to achieve stability of secure information systems as well as to fi ght against cyber crime.
In regular crimes, judges normally
understand the evidence of the
crimes. However, digital evidence
are difficult to be evaluated.
- [Digital Forensics]
- Today many of evidence of frauds and crimes remain on various digital equipments in the form of digital data. But those digital evidence is quite fragile as well as easily erased and tampered. It is also quite difficult to understand and investigate to person other than professional engineers. Digital forensics is a set of related technologies to aquire and investigate those digital evidence which are related to cyber crimes , frauds , system incidents and failures. It is a new field of research and we are developing a variety of techniques as a pioneer of the digital forensics.
The Integrated Systems Laboratory researches and develops practical hardware and its design methodology for information devices and systems integrated on VLSIs.
- Reconfigurable Hardware
- Reconfigurable devices which can be programmed their functionality after fabrication are now widely used in various areas, including spaceships and satellite systems in which capability of remote maintenance is highly demanded. We are developing reconfigurable devices with higher level of reliability against soft-errors caused by cosmic rays in satellite orbit.
- Ultra-low-power System
- Mobile information devices such as cell phones, PDAs, and portable game consoles work with battery power, and thus reduction of power consumption is desired to achieve longer operation. Extremely low-power design is essential in devices supplied by solar cells such as sensor nodes. We are developing low-power architectures and circuits for systems that operate with energy harvested by solar cells.
Design Automation Tool
for Reconfi gurable Device
Chip Micrograph of Solar Cell
Fabricated on CMOS LSI
MxD lab are exploring new experiences provided by the media technology for contributing to our real life. Our research target covering all fields such as daily life, education, transportation, medical care, games. Starting from actually observation and understanding the our activity, we try to find essential problems at the real site and then solve them with media technology.
The e-Society Laboratory is a multi-disciplinary, international research and learning space centering its work around individuals, social groups, organizational networks as well as public administrations in the age of digitalization. Classical IT projects are envisaged in a broader environmental and social perspective encompassing the exchange with experts from architecture, education, linguistics, sociology and cognitive science, but also from engineering and physics. One of the distinctive features of the laboratory is its international character. The working language is English. Student members of the Laboratory are typically expected to report results of their work at professional meetings, including international conferences, while graduate research projects are to produce results that would be reported in international journals. Graduate students' work is coupled with external international affiliates of the laboratory that include research groups from the University of Zurich, the Techinical University of Cyprus, and the University of Oxford.
Extracurricular laboratory events include birthday cake-parties, cross-cultural BBQs and city tours.
As human society is getting highly-developed and complicated, it becomes hard to design suitable social systems for the society. Social simulation has been regarded as a promising method to understand complicated human society and predict probable social problems. The research objective in this laboratory is to explore technologies to design systems, mechanisms, services for human society in the future. The core technology is Multiagent-based simulation (MASim) which has been focused as a promising form of micro-simulation because it is suitable to reproduce complex phenomena in human societies.
We have conducted MASim in various domain such as traffic, energy, disasters, and internet communities. Especially, in the traffic domain, we have developed massively multi-agent simulator on which each agent can make its own decision every few seconds. Then, we successfully reproduced traffic flows under conditions applied for social experiment held in Kyoto City in 2007.
Traffic Simulation In the heart of Kyoto City
Simulation is not a technology to accurately predict the future, but a tool to find or point out possible form of the future. We will try to develop MASim-based technologies which can promote collaborations among humans and machines/softwares to solve diverse issues in human societies through interactions among them.
The brain is considered to use relatively small amount of energy for its efficient information processing. The goal of our
researches is to clarify the principle of information processing in the brain: a flexible and energy-saving computer with low power consumption. In particular, we focus on stochastic information processing on a characteristic architecture “cortical local circuit”, which is commonly observed among various cortical areas.
To this end, we construct electrical measurement systems for cortical activity, record neural activities by physiological experiments, analyze the data using statistical techniques, construct mathematical models, and extract rules for cortical information processing and nonlinear dynamics. We expect taking various approaches to “cortical local circuits” gives new perspective to physics, engineering, and neuroscience.
- [Current research interests]
- * Stochastic information processing by neurons in the cortical local circuit
* Nonlinear dynamics of neuronal population
* Information analysis of physiological experimental data
* Statistical properties of dynamical systems of complex network
In the Media Sensing Laboratory we are conducting research on sensing technology and the related signal reconstruction techniques in various areas such as medical imaging, image processing, audio signal processing, and radar signal processing.
For example, if we need images of inside of a human body, we use a magnetic resonance imaging (MRI) or an X-ray computed use a high-speed camera. We are developing technologies that enhance capabilities of such devices. Specifically, we are interested in how to reduce the scanning time in MRI and the radiation exposure in CT without loss of image quality. How to keep the number of pixels per frame, as many as possible, of a high-speed camera is also our research topic. Recently, we are doing research on sensing technology about audio and radar signals.
- [Medical Imaging]
- ・image reconstruction in high-speed MRI and low-dose CT
- [Image Processing]
- ・improvement of high-speed camera and light field camera
- ・single image super-resolution via neural network
- [Audio Signal Processing]
- ・freshness estimation of vegetables using sweep sound
- ・blind sound source separation using several microphones
- [Radar Signal Processing]
- ・data processing for phased array weather radar
High speed MRI
high speed cameras
Freshness estimation of
tomatoes using sound
Data processing for phased
array weather radar
In recent information society, it is necessary to use diverse knowledge and information in different areas, organizations, and disciplines. Our research aims to facilitate access to knowledgeーsupported by computersーand its semantic-based use of knowledge computing. The key technologies cover “ontology engineering,” i.e., semantically defined concepts, and “semantic technology,” i.e., interlinking heterogeneous resources on the web. In previous research, one of the implemented systems included ontology-based procedural activities to describe nurses’ expertise, developing a tablet- computer application for education and training of novice nurses. This system has successfully been used in real-world applications. Another research field is based on epidemiological modeling and simulation of knowledge propagation in societies. Further research includes the analysis of color representation in different cultures. This laboratory conducts fundamental research, development of application systems, and research applied in the real-world containing a variety of disciplines and domains.
Tablet-based learning application software
for nursing activities
A situation of its real use in a hospital
An IoT (Internet of Things) is the network consisting of not only computers/smartphones but also various physical objects (or things) that have connectivity to the Internet. These objects can be embedded with various devices and electronics (e.g., sensors, robots, cars) in the real world, hence the IoT is attracting attention in research and business fields for constructing new network services. In our laboratory, we are focusing on greater value that can be achieved by interconnecting things through the Internet, and are researching and developing enabling technologies for achieving the innovative network system which can improve quality of life in the real world.
- Topic 1: Ecology and environment observation system using embedded system
- The embedded system is a small-sized, low-power, low-cost computer which can include various sensing capability and network interfaces. By utilizing the embedded system, we are researching and developing ecology and environment observation system, for example, snowfall measurement system in an area of heavy snowfall (Figure 1), and seabird observation system in an isolated island.
- Topic 2: Information transmission system using air vehicles
- Recently, a high performance but small-sized air vehicle (e.g., quadrotor helicopter, or drone) which is equipped with the wireless communication and auto-pilot functions has become widely available. By utilizing the air vehicle, we are studying a new information transmission system that the air vehicle stores, carries and forwards the data even in a dangerous field (e.g., disaster area, field of ecology/environment observation) where the network infrastructure cannot be easily constructed (Figure 2).
Real-time measurement system of snowfall
Information transmission system utilizing air vehicle
Affective Engineering and Computer Arts Lab(AECAL)researches and develops models to infer human affective, emotional, or sensual responses and applies these models to specific objectives. When a person is exposed to a stimulus, the experience may be associated with quantifiable internal and external responses. Because the objective systems are to satisfy specific, practical objectives, the models applied range from statistical models to methods of soft computing and intelligent systems (also known as artificial intelligence, or AI). AECAL develops new methods to meet those needs.
There are many areas where affective responses are central to the fuctioning of essential systems. In security, for example, user feelings lead to behaviors that may either protect or endanger data, systems, and people. In education, how students feel has a direct influence on how they. In healthcare, patients feel is also a major final goal. In these and other areas, intelligent systems are descovering novel solutions to problems once considered beyond the scope of engineering.
AECAL also applies affective engineering and intelligent systems to the arts. Computers have often been tools for the arts but, more and more, intelligent systems are becoming full partners in the creation process in every day. Computer arts today will become the innovative technologies of tomorrow because the arts have no borders.
The research directions of this laboratory are in the fields of robotics and biological systems that demonstrate aesthetically "beautiful" agile movements and provide a motivation for developing dexterous intelligent machines and robots. While living creatures demonstrate motion intelligence, our understanding of their motor behavior is not complete. On the other hand, we possess a much better understanding of how modern robots work, although they clearly have shortcomings. One of the paths toward bridging the gap between robotic and biological systems is in designing intelligent machines and control systems, based on biologically-inspired principles, classical mechanics, and control theory.
- [Biological Motor Control]
・Mathematical analysis and modeling of natural human movements
- [Robot Dynamics and Control]
・Dynamic models and motiion planning for underactuated and non-holonomic robotic systems
・Self-organization, learning control, and emergence of motion patterns
・Biologically-inspired principles, computational models and control strategies of force and stiffness distributiion in cooperative movements
・Human-robot interaction, haptic interfaces, and acquisition of skillful movements in artificial dynamic environments
・Internal forces, stability, and control strategies for cooperative movements by muscle-like parallel mechanisms and multi-fingered robotic hands
・Optimal control-based algorithms for motion and force redundancy resolution
・Driving principles and gain control of rolling robotic systems
Our laboratory research deals with human-oriented modeling and development of visualization systems using VR tools, sensory devices, and experimental data bases. As a basis for the visualization systems, we develop real-time algorithms and methods for gridding, volume rendering, 3D segmentation, mesh reconstruction, and morphing, which are specifics for human-oriented modeling. We conduct body shape modeling using Digital Human manikin models and 3D full-body scan data. Modeling of body shapes can be done for the scenarios of weight gain/loss, muscularity gain, and effects of ageing, which are important for healthcare and beauty services. Combining with motion capture(MoCap) data, 3D scans are used to estimate dynamic parameters, necessary to realistically model human movements. Digital Hand models are used for ergonomics factors estimation, as such as grasp quality and stability. We pay attention to visualization systems with haptic devices for VR surgery and VR nursing. Digital Human modeling is driven by real-time sensory and feedback devices(trackers, accelerometers, haptics), and utilizes experimentally collected data bases(3D scans, CT/MRI data). We also conduct experimental processing of data collected from haptic devices to model and predict human hand movement in constraint dynamic environments and to study human balancing and reaction skills.
With the advent of mobile, wareable, and ubiquitous devices, data on human activities, situations, and physical conditions can be collected. Systems and services that utilize the results of analyzing such data provide advanced functions that have been difficult to achieve so far and support tasks performed by manual work, human intuition, and human wave tactics.
These technologies also improve the human capability, i.e. human augmentation, and contribute to the construction of new social structures, such as health care, medical support, business support, sports, entertainment, interface, and authentication.
In our laboratory, we develop technologies that enable computers to understand human, to transfer information appropriately to people, and to efficiently manipulate computers by sensor information processing, machine learning, cognitive science, usability, system design and so on.
Social problems increasingly become too complex to be solved within one discipline. To address such problems, it is necessary to collect intelligence from a whole society. Our laboratory aims at forming social intelligence based on services computing technologies. Especially, by interconnecting web services of human/machine intelligence, we are developing a service-oriented platform that accelerates the construction of social intelligence.
For example, to overcome language barriers caused by globalization, we are operating the Language Grid, a platform to customize a multilingual service according to a community’s context by combining various language resources. Also, to save endangered languages, we are driving Indonesia Language Sphere project that comprehensively creates bilingual dictionaries of lowresourced languages by combination of automatic generation and multiethnic crowdsourcing.
For the objective of realizing an intelligent society that is human-friendly and of high human affinity, it is important to make computers to be able to automatically acquire knowledge from images/videos and understand their meaning. Our laboratory works on both the theoretical research topics and practical applications of these technologies. In particular, we focus on the research on human detection, human re-identification, fine-grained human attribute classification (age, clothes, physique, belongings,etc.), and human action recognition. On the basis of these key technologies, we attempt to analyze, recognize and predict the events in the images/videos at a higher level. We pay special attention to the end-to-end strategy (directly obtain wanted information from raw data) using deep learning.
Examples of our research topics: Fine-grained human attribute classification, robust pedestrian detection and recognition against occlusion, fine-grained action recognition, video-based first impression evaluation, automatic video digest generation, automatic rating on violence level of videos, etc.