Basic Sciences Category

Prize Field : Mathematical Science

The 2006 Kyoto Prize Laureate in Basic Sciences

Dr. Hirotugu Akaike
Born: November 5, 1927 (Age 78)
Nationality: Japan
Occupation: Statistical Mathematician
Professor Emeritus, the Institute of Statistical Mathematics

Received a B.A. (1952) and a Ph.D. (1961) from the University of Tokyo. Started his career at the Institute of Statistical Mathematics, Japan, in 1961, serving as Director General of the Institute from 1986 to 1994; appointed as Professor Emeritus at the Institute in 1994. Accolades include receiving the Okochi Memorial Technology Award (1980) in the field of industrial engineering and production technology; the Asahi Prize (1989); the Japanese Medal with Purple Ribbon (1989); and the Japan Statistical Society Award (1996).

A statistical mathematician who made a major contribution to statistical science and modeling by developing an information criterion known today as the "Akaike Information Criterion" (AIC).

In the early 1970s, Dr. Hirotugu Akaike formulated the Akaike Information Criterion (AIC), a new, simple and highly practical criterion for the selection of statistical models. In so doing, he established a new paradigm bridging the worlds of data and modeling, thus contributing greatly to the information and statistical sciences.

Background
Rapid advances in science and technology in the 20th century brought numerous benefits to society, but at the same time exposed many new problems. Furthermore, in the 21st century, rapid globalization and "informatization" have resulted in the development of strong global links that have transformed the world into a huge network of mutually dependent systems. Consequently, it is no longer possible, in many cases, to solve problems within the framework of a single isolated system; it is instead necessary to grasp, analyze and forecast problems in the context of this global network of closely linked systems. Modern statistical sciences are expected to help our understanding of the study of such complicated, uncertain and dynamic systems, or the study of situations where only incomplete information is available. The main role of statistical sciences must be to give useful scientific methodologies for the development of new technologies and for the further development of society, which is characterized by increased uncertainty. One of the characteristics of the statistical sciences is their interdisciplinary nature, with applicability in various fields. For example, the role of the statistical sciences has recently become increasingly important in the understanding and forecasting of phenomena in economic-related fields, such as finance and insurance; safety-related fields, including pharmaceuticals, food and transportation; natural phenomena, such as weather, natural disasters and the environment; and in the management of huge systems.

Dr. Akaike's Achievements
Starting in the early 1970s, Dr. Akaike explained the importance of modeling in analysis and in forecasting. He formulated the Akaike Information Criterion (AIC), which facilitates selection of the most appropriate model from a number of different types of models. Ever since, the AIC has been exerting a powerful influence on the development of the information and statistical sciences.

In order to understand and forecast phenomena from a vast quantity of data obtained in experiments or observations, it is necessary to construct a hypothetical statistical model. The selection of such a model is highly subjective, as it is made on the basis of a researcher's own ideas, knowledge and experience. Therefore, it is essential to estimate the most adequate model among the possible candidates. However, from a practical standpoint, this was very difficult because of the finite number of data and the lack of an objective criterion for selection. The AIC offers a solution to this problem, which seems to be common in almost every field of engineering and science. Consequently, the role and meaning of the AIC as a criterion for estimating statistical models have become extremely significant in the development of statistical science. The AIC is built into commercial statistical software packages, and is also widely used in such diverse areas as gene analysis; image compression technologies; and vehicle stability control technologies, among many others.