Message from Director-General

Tomoyuki Higuchi

November 21, 2016

Tomoyuki Higuchi

Director-General
The Institute of Statistical Mathematics

The previous fiscal year was one in which the subject of artificial intelligence(AI) created media headlines seemingly every day. In particular, the victory of the Google-developed AlphaGo AI system over an international Go master brought news of rapid progress in AI technology to the general public in a way that was particularly easy to understand. Of course, it was already some 20 years ago, in 1997, that an international chess grandmaster was defeated by the Deep Blue system developed by IBM. The dramatic transformation of the AI environment over the ensuing years can be attributed above all to one single factor: the explosive growth of data. For example, whereas the quantity of genomic sequence data that can be read in one second has increased roughly a millionfold over this interval, the speed of computation — as characterized by Moore’s law, the empirical formula describing the evolution of computer performance — has grown “only” by a factor on the order of ten thousand. This data explosion is the true breakthrough enabling modern AI technology.

The impact of AI is evident across a vast range of the modern world — from our daily lives in the world around us, to an industrial restructuring that might be termed a fourth industrial revolution, as well as to the front lines of research and development in science and technology — and strategies for making effective, timely, and appropriate use of rapidly growing quantities of data will be the keys to success in the global society of the future. The subject of machine learning, which is responsible for the core functionality of AI systems, shares many aspects of its conceptual structure and theoretical foundations with statistical mathematics. For this reason, four years ago the Institute of Statistical Mathematics (ISM) created the Research Center for Statistical Machine Learning, which quickly began making contributions to relevant fields. This Center is one of five centers promoting the NOE (Network Of Excellence) Project. Moreover, because Japan faces an overwhelming shortage of domestic data-handling professionals, the School of Statistical Thinking was established around the same time as the Center; the establishment of this School demonstrates our focus on training new generations of data scientists by promoting the Project for Fostering and Promoting Statistical Thinking.

During the period covered by the Second Midterm Goals and Plans, which ended in March 2016, we are proud to say that, supported by the dual pillars of our NOE Project and the Project for Fostering and Promoting Statistical Thinking, we accomplished our Second Midterm Goals and more: designing a research network, training young personnel via OJT, advancing the Project for Fostering and Promoting Statistical Thinking, and providing resources for updated and new supercomputers and High Performance Computing Infrastructure (HPCI), among other accomplishments. Moreover, the previous fiscal year was one that witnessed a further entrenchment of our industry/government/academia collaborations. We completed an agreement with Tachikawa City concerning collaboration and cooperation, and participated in the Tachikawa Residents Survey; by assisting with events sponsored by the city, we were able to contribute to the region and to enhance our collaborative relationship with it. In April 2016, the ISM officially entered the period covered by the Third Midterm Goals and Plans. In addition to fostering research activities that encompass both pure and applied work while maintaining a keen awareness of connections to the real world, our objectives include sharing the fruits of our research with industrial, governmental, and academic institutions, as well as local communities throughout Japan, and strengthening the functionality of institutions.

One sometimes comes across news stories rosily predicting that AI will soon solve all societal problems, but the methodology and techniques of today’s AI will never lead to solutions for some difficult social problems, such as predicting natural disasters and other massive risk incidents, supporting mental-health care, and offering real, not just virtual, replacements for the knowledge and skill of seasoned professionals in complicated manufacturing processes. We are dedicated wholeheartedly to building a laboratory — based on the foundation of “intelligence” at which statistical mathematics excels, including prediction, discovery, optimization, and control — to meet the needs of academia and society. We ask for a continuation of the support and understanding you have shown the ISM, and, more than ever, for your valuable advice and guidance as we proceed with this endeavor.

Backnumber