Laboratories – Informatics

Mathematical Informatics

String Data Processing Laboratory

String Data Processing LaboratoryMembers : Prof. Masayuki Takeda / Assoc.Prof. Hideo Bannai / Assoc.Prof. Shunsuke Inenaga /Asst.Prof. Yuto Nakashima

keywords : Algorithms, Data structures, Data Compression, String Combinatorics

In computers, all information is represented as a sequence of characters or “strings.” With our signature phrase “Everything is String” our laboratory studies matching/compression/searching/learning/discovery on string data. Through discovery of novel mathematical characteristics of strings allowing increased data-processing speed and reduced memory usage, we aim to develop not just evolutionary but revolutionary technology. We believe it is most necessary to build timeless and universal fundamental theories not swayed by short-sighted or short-lived applications.

Mathematical Engineering Laboratory

Mathematical Engineering LaboratorMembers : Prof. Jun'ichi Takeuchi / Assoc.Prof. Yutaka Jitsumatsu

keywords : Machine Learning, Minimum Description Length (MDL) Principle, Stochastic Complexity, Information Geometry, Information Theory

Our laboratory aims to find out mathematical structures of various problems in computer science and digital communication and to derive universal solutions based on the mathematical structures. Our research topics include basic theories such as learning theory, machine learning, information theory, information geometry, communication theory, network theory, nonlinear system theory and their applications. Specific applications are cyber-attack detection on the Internet, super resolution, pattern recognition, CDMA communication, analog/digital conversion, and error correcting codes. Through these researches, we develop human resources who are responsible for the fundamental technologies in advanced information society in future.

Algorithm Theory Laboratory

Algorithm Theory LaboratoryMembers : Assoc.Prof. Shuji Kijima / Assoc.Prof. Yukiko Yamauchi

keywords : Randomized algorithms, Discrete mathematics, Mathematical programming, NP-complete, Stable marriage

Algorithm Theory Laboratory is widely interested in the principles of computing, particularly algorithm theory. For a bunch of problems originating from the real world or motivated by theoretical computer science, our research interest includes design of algorithms and mathematical analysis from the view point of correctness, efficiency, robustness, etc. Assoc. Prof. Kijima is mainly involved in the topics such as randomization and derandomization including Markov chain Monte Carlo (MCMC), approximation algorithms, discrete mathematics including graph theory, matroid system, submodular functions, etc. Assoc. Prof. Yamauchi is mainly involved in a variety of topics on distributed coordination, such as fault-tolerance of distributed systems, self-organization of autonomous mobile robots, and game theory in distributed environment.

Information Security & Multimedia Security Laboratory

Information Security & Multimedia Security LaboratoryMembers : Prof. Kouichi Sakurai / Asst.Prof. Wissam Razouk

keywords : Network Security, Security Camera, Security Robot, Adversarial Machine Learning, Computer Security, Cryptography

Nowadays, not only people but things are getting increasingly interconnected in what is called Internet of Things (IoT). In a world where almost everything is connected, if an attacker gets control of one of these networks it can be disastrous. Attacks can go as far as changing the election results of a country (USA’s 2016 election was strongly influenced by Russian cyberattacks). Moreover, in a recent report from Forbes, cybercrime is projected to reach 2 trillion dollars by 2019. To protect society, we research new technologies and paradigms for security related applications.

Laboratory of Intelligent Systems

Laboratory of Intelligent SystemsMembers : Assoc.Prof. Danilo Vasconcellos Vargas

keywords : Deep Learning, Neuroevolution, Action/Image Recognition, Multi-agent based Intelligence, Bioinspired Artificial Intelligence, Artificial General Intelligence, Evolutionary Computation, Reinforcement Learning, Adversarial Machine Learning

In the Laboratory of Intelligent Systems we create novel AI engines as well as build robust and adaptive intelligent systems. Current AIs can solve 19x19 versions of Go but behave poorly on easier 9x9 versions of the same game. Similarly, image recognition algorithms can reach 96% accuracy (supra-human) on tests and be fooled by only one pixel change. In other words, current AI lacks the robustness and adaptation present in even simple living beings. AI is based on engines that allows it to learn and reason over things, this lab builds novel engines based on different paradigms to reach high levels of robustness and adaptiveness intrinsically. Interestingly, by increasing the robustness and adaptiveness, other problems like Transfer Learning, One-Shot Learning would also be solved at the same time, igniting, possibly, a new age of intelligent systems.

Theory of Computing Laboratory

Theory of Computing LaboratoryMembers : Assoc.Prof. Akitoshi Kawamura

keywords : Mining, Information Retrieval, Computer Simulation, Database

  1. Algorithmics
  2. Computation Theory

Intelligence Science

Cognitive Science Laboratory

Cognitive Science LaboratoryMembers : Prof. Shuji Mori / Prof. Kazunori Shidoji / Asst.Prof. Nobuyuki Hirose

The cognitive science laboratory explores functions of human mind for their engineering applications. Prof. Mori investigates auditory temporal resolution and categorical speech perception through behavioral experiments and functional brain imaging measurements and attempts to develop new hearing tests, using a variety of psychophysical techniques. Prof. Shidoji focuses on estimation of driver's state in driving simulator and real-road driving, its application to development of automated driving system and driver support system, and perception and cognition in virtual reality environment.

Data Mining Laboratory

Data Mining LaboratoryMembers : Prof. Einoshin Suzuki / Asst.Prof. Tetsu Matsukawa

keywords : Data mining, Machine learning, Autonomous mobile robot, Robot, Deep learning, Anomaly detection, Exception discovery, Classification

In data mining, which aims at sophisticated discovery of potentially useful and understandable patterns from massive data, we tackle diverse issues from fundamental ones to applications with various bases including machine learning. Examples include data processing such as data squashing and data structure, pattern discovery such as various types of exceptions and rules, pattern interpretation such as information visualization and human factors, and other issues such as problem formalization. Moreover we conduct various kinds of research including autonomous mobile robots using machine learning and data mining techniques as well as deep learning on image, video, and text data.

Machine Learning Theory Laboratory

Machine Learning Theory LaboratoryMembers : Prof. Eiji Takimoto / Assoc.Prof. Kohei Hatano

keywords : Online decision making, Computational learning theory, Computational compelxity

The problem of decision-making by predicting future data from the past arise in many applications such as stock investment, item recommendation, routing, updating kana-kanji conversion dictionary, and so on. Our group is trying to develop ingenious methods of decision-making for various problems by using machine learning techniques. On the other hand, we also apply the methods developed to optimization problems in machine learning. Furthermore, for various classes for knowledge representation such as Boolean circuits, decision diagrams, neural networks, comparator networks, we investigate their mathematical properties and relationships between them, thereby we analyze computational efficiency of decision making methods.

Automated Reasoning & Applications of Machine Learning Laboratory

Automated Reasoning & Applications of Machine Learning LaboratoryMembers : Assoc.Prof. Hiroshi Fujita / Asst.Prof. Miyuki Koshimura

Automated Reasoning is a key technique in intelligence science and technology. It is expected to be applied to a wide range of fields including hardware/software verifications, supporting legal reasoning, and so on. Currently, we are developing efficient automated reasoning systems such as SAT/MaxSAT solvers and their applications. We have tackled the following problems including large-scale combinatorial problems: Ramsey problem, scheduling problem, correcting errors in AES key schedule images, coalition structure generation problem, and inductive logic program.
Nowadays, various data are collected every day. Many machine learning techniques have been developed so far. We would like to practically show the effectiveness of these techniques using real data. Currently, we are working on the following two applications: classification of several objects in microscopic images of body fluids and anomaly detection for factory equipments.

Multi-Agent Laboratory

Automated Reasoning & Applications of Machine Learning LaboratoryMembers : Prof. Makoto Yokoo / Asst.Prof. Taiki Todo

keywords : Market Design, Artificial Intelligence, Mechanism Design, Matching, Combinatorial Auctions, Repeated Games, Prisoner’s Dilemmas, Hospitals/Residents Matching, POMDP, Constraint Satisfaction

The main research field in our laboratory is multi-agent systems, where multiple intelligent agents coexists. Especially, our research focuses on systems where humans and software agents interact and coordinate. Specific research topics include two-sided matching and auctions, for which we model agents’ behaviors based on game theory and micro-economics, and develop/analyze social decision rules based on algorithm theory and optimization.

Neuroimaging and Neuroinformatics Laboratory

Neuroimaging and Neuroinformatics LaboratoryMembers : Prof. Keiji Iramina

keywords : Neuroscience, Neuroengineering, Brain Information Science, Event related potential, fMRI, Brain Machine Interface, Cognitive function, Mild cognitive impairment (MCI), Alzheimer disease

Iramina’s lab is under the administration of the Faculty of Information Science & Electrical Engineering, Kyushu University, and Graduate School of Systems Life Sciences which is a unique educational organization.

There are two major research fields in our lab. One is brain function imaging which aims at the elucidation of human brain function; the other one is brain function modeling which is applied to various fields by constructing the model of brain activation. In details, we study in the fields of the measurements of brain function by EEG (Electroencephalography), NIRS (Near-Infrared Spectroscope) and TMS (Transcranial Magnetic Stimulation), the development of measurement technology and the simulation of brain activation. The elucidation of the mechanism of brain function is one of foundations of life science, and it can be applied to almost all the fields. Have a deep understanding of brain information processing, and apply the research results to fields of life science, medicine, welfare and education is the purpose of our study.

Since we are studying in an interdisciplinary domain, we take into account the collaboration of medicine, biology, pedagogy and psychology is important in our study.

Natural Language Processing

Natural Language ProcessingMembers : Prof. Yoichi Tomiura

keywords : Machine Learning
Organizing Information, Statistical Model, Deep Learning, Natural Language Processing, Text Mining, Data Mining, Data Science

Natural Language Processing (NLP) is a field on technology to process sentences written in natural language such as Japanese and English using computer. As informationization advances and a large amount of information is flooded, NLP focuses attention as a technology for efficiently accessing necessary / important information and for analyzing a large amount of text. With the advent of Deep Learning, the performance of various NLP technologies including machine translation has been remarkably improved, and expectations for NLP are increasing more and more. We are conducting research on identifying and clustering sentences or documents based on parameter estimation of statistical language model, and research on estimating similarity between sentences or documents by Deep Learning.

We are also conducting research on the analysis of olfactory information using a model similar to the statistical language model used in the above research. Based on the images of the activation patterns of the neurons on the olfactory bulb (the first brain part receiving the odor information) of the rats when smelling various substances and the physical and chemical properties of the substances, we are working to identify the primitives of odors and the parts of the olfactory bulb that ignites when they are detected. In addition, we are working to separate and visualize odor traces (odor source) based on measured data by multi-channel odor sensor.

3D Multimedia Contents Laboratory

3D Multimedia ContentsMembers : Prof. Yoshihiro Okada

keywords : 3D-CG, Multimedia, HCI

Our laboratory is researching and developing fundamental technology for 3D multimedia contents of still images, videos, 3D shapes, motion data and so on. In addition to search and creation technology for them and visualization technology, the research interests of our laboratory also include voice input/output interface for 3D-CG contents, motion input interface based on video images, virtual reality applications using a haptic device like Phantom, network collaboration technology for instantly and easily creating a virtual space of 3D-CG in which multiple users can take various intellectual activities collaboratively with each other. Our laboratory also conducts research on the development environments of 3D games and educational materials using recent ICT.

e-Science Laboratory

e-Science LaboratoryMembers : Assoc.Prof. Daisuke Ikeda

keywords : computational complexity, mathematical logic, foundation of mathematics, numerical analysis, validated numerics, differential equations, computation model, randomness

Due to big data and the development of ICT, computer simulation and data analysis with computers are used in many disciplines. While computers have been supporting tools for experts, there is an emergence of a new discipline, called e-Science, in which computers are main approaches.
In our lab, under the vision that "general public will be participating a process of science", we are conducting researches, such as computer simulation and data mining, and infrastructures for e-Science.

Statistical Learning Laboratory

Statistical Learning LaboratoryMembers : Assoc.Prof. Hiroto Saigo

keywords : bioinformatics, cheminformatics, machine learning, statistics, data mining

Our primary research interests is in the development of statistical learning methods, which gives foundation for data science and deep learning. Due to the recent public interests in artificial intelligence and machine learning, various industries are seeking a way to make good use of it. In solving real-world problems, however, what is required for data scientists is not only to have deep understanding on various machine learning methods, but also to become familiar with the domain of the facing problem. In this regard, we put an emphasis on dealing with real-world data, and always use it for evaluating proposing methods.
One of the characteristics of this group is its focus on biology and chemistry, such as developing methods for handling genes and chemical compounds, however, our research interests is not limited to these areas.