ホーム

MEMBERS | Program Members

Sadanori Konishi

MEMBERS

Message Download Consortium

MEMBERS

Program Members

New Development in Statistical Science

Sadanori Konishi
Degree: Doctor of Science (Hiroshima University)
Research Interests: Statistical Science
Unit:Uncertainty

Report

In the advanced information technology environment of the 21st century, various science and technology fields, such as life science, information technology, earth environmental science, systems engineering, and financial engineering, have made a great strides beyond initial predictions, and new academic fields are continuously being established. Mathematics as a science has directly and indirectly supported these developments as new research activities have been actively pursued using innovative ideas in an advanced computing environment. In particular, achievements in mathematical science, statistical science, and computer science, have provided mathematical and statistical techniques to meet the challenges of extremely difficult problems.

In fact, in modern society, highly advanced computer systems and advances in electronic measurements and measurement technologies are combined to store large amounts of various data from all areas of sciences and industry, including data to characterize genes such as microarray data in life science, remote sensing data transmitted in real time from an Earth observation satellite, continuous data of phenomenon process and operation process in systems engineering, fluctuating economic data such as stock prices and foreign exchange rates accumulated at a high frequency, and Point Of Sales (POS) data, which is read and collected using optical scanners from barcodes on merchandize. This vast quantity of data is being accumulated and organized as databases. Hence, it is desirable to research and develop new statistical data analysis techniques to efficiently extract and process useful information as well as elucidate patterns hidden within the data in order to analyze various phenomena and to discover their governing laws.

Our global COE, Education and Research Hub for Math-for-Industry, tackles research and development of a new modeling method to understand complex non-linear phenomena, which are being researched around the globe as a method in the advanced computer age, thereby providing tools and technologies useful for solving problems that various sciences and industries face. In fact, modeling of an object is indispensable for extracting information from a large amount of data, for discovering knowledge, and for understanding, predicting, and controlling complex natural and social phenomena. Additionally, in this modeling process, selecting an optimum model is an important key to appropriate logic and elucidation of knowledge because this allows us to analyze complex and varied data, such as graphic information, real-time recorded data of the motion process of human walk, and genome data in life science, and to effectively perform modeling based on three-dimensional data of protein.

To date, among statistical methods proposed by Japanese researchers, which have been applied to natural sciences as well as social science fields and have greatly contributed to understanding phenomena and discovering knowledge, a notable method is a model evaluation standard called Akaike’s Information Criterion (AIC) (Dr. Akaike, who proposed the AIC, received the 22nd Kyoto Award in 2006). AIC has been playing a major role in understanding phenomena in various fields of science, including controlling boilers at thermo-electric generation plants, designing automatic navigation system of ships, and analyzing time series data in economics.

Recent rapid progress in electronic measurement and instrumentation technologies has enabled a wide variety of data, including genome data, real-time record data of operation processes and dimensional and three-dimensional data, to be collected. Hence, it is imperative to conduct research and development on new modeling methods in order to extract useful information and patterns with a high efficiency, especially for non-linear modeling methods to capture complex non-linear structures.

Our Global COE tackles issues directly related to real-world phenomena while simultaneously strengthening collaborative ties with various other branches of science and industry. Through this approach, we strive to conduct research and development of novel statistical data analysis technologies to effectively investigate complex non-linear modeling by integrating knowledge in statistical science, mathematical science, and information science. Moreover, we aim to understand and recognize the needs of society, and to apply these needs to our education and training of graduate students as well as young researchers at the Graduate School of Mathematic at Kyushu University. Hence, we strive to create innovative research results in novel statistical sciences with regard to these research topics.

RETURN LIST