CONCLUDING REMARKS

From a very early age, Japanese school children are frequently reminded of their country's vulnerable economic condition. They live in an island nation, with a population equal to a third of the United States crowded onto a land mass slightly larger than California, and of which almost three-fourths is mountainous and essentially uninhabitable. They have no natural resources and not enough space to grow their own food. They have been grappling with the related problems for several generations.

Their involvement in the Second World War was largely an act of desperation, to increase their space by acquiring part of China. When that failed they determined that their only option was to succeed in the international market place, earning enough money to buy what they need to survive. Their strategy since that time has been to import raw materials and turn them into products that can be sold abroad.

With respect to their now taking a leading role in fuzzy systems research, it has been proposed that this is because Japan is one of the Eastern cultures, which are reputed to be more receptive of vagueness and imprecision. My personal opinion is that this is only a small, and perhaps negligible, part of the overall cause. It seems rather that Japan's investment in fuzzy systems is only one of many instances of the same overall strategy. They constantly scour the horizon, watching for new scientific ideas, which in effect are regarded only as other raw materials, items waiting to be transformed into marketable commodities. This approach to technological progress is evident even in their policy toward funding research at large corporations and in universities. The prevailing emphasis has been on applications rather than theory. A nice illustration of this fact is the following breakdown of presentations at the congress of the International Fuzzy Systems Association (IFSA-91) held in Brussels, Belgium, last July. The papers were subdivided into four areas: Artificial Intelligence (AI); Engineering (Engin); Computer, Management, and System Sciences (CM&SS); and Mathematics (Math). Note not only the distribution but the proportion of total participation by Japanese.

Total By % Topic Papers Japanese Japanese

In order to accomplish this it has been necessary for them to excel in business and technology. They are a people who have no choice but to work hard and live by their wits. In order to accomplish their goals they are accordingly inspired to acquiesce in a rather tightly controlled and conservative society, with a high value placed on social harmony. The Japanese are well aware that they cannot permit excessive squabbling among themselves, for they otherwise cannot live together in such a small space and probably could not collectively survive. This feeling inspires in them a desire to care for, All and cooperate with, one another in ways mostly unthinkable in the West. It also allows them to respond rapidly and in unison to meet new challenges.

Against this backdrop we have seen this nation rise rapidly to the forefront in the realm of information technology.

ΑΙ

Engin

CM&SS

Math

The strong emphasis on applications has been orchestrated largely by MITI and STA through their funding policies. In contrast with the United States, MITI can to a certain extent dictate where large corporations will

focus their R&D, simply by creating funding opportunities in desired areas while withdrawing funding in others. They are also not afraid to invest occasionally in high-risk ventures.

Just as in the United States, however, the granting agencies do make their decisions on the basis of advice received from their leading scientists. And this points to another possible reason why fuzzy systems have flourished in Japan. Japanese university professors are all paid on a 12-month basis, so that they have no need for summer salaries and are typically provided at least the minimal funding necessary to do their research. Hence, there is not as much incentive as in the United States for faculty to compete for research funds, which means there is less cause to discredit ideas other than one's own. Discussions regarding the pros and cons of a new idea can take place in a much more friendly and collegial manner, simply because one's livelihood and/or academic survival is not so much at stake. When a few Japanese professors took an early interest in fuzzy systems, they were at least treated with respect, if not actively encouraged.

Fuzzy systems research has been suppressed in the United States largely because the leading AI figures, and principal advisors to the granting agencies, come from the traditional background of symbolic approaches to AI. Accordingly, the models of semantic inference proposed by Zadeh seemed not only strange but also threatening. Out of this came an apparent (although unstated) policy to provide only minimal funding in this area. Also, because U.S. academics do not value social harmony and mutual respect as much as do the Japanese, there was great sport made in some circles of ridiculing and alienating anyone who was identified with the fuzzy systems movement. This only exacerbated the problem of gaining support.

As to what might have been done, or what could be done in the future, to prevent similar phenomena from occurring, I can only offer a few tentative suggestions. There is, of course, little that can be done about the lack of importance placed on values like social harmony and mutual respect, but then this may not be altogether necessary. Out of rugged individualism and competitive self-promotion comes also a wealth of both scientific, philosophic, and artistic creativity. Indeed it is still the case that the deep theoretical breakthroughs are being accomplished in the West. This is because we continue to support the intellectual infrastructure necessary for this to occur. What is needed, therefore, is really only a better system for (1) protecting and encouraging fledgling ideas and (2) transforming theory into applications.

For the former I would propose a partial reorganization of the NSF, creating a branch organized somewhat along the lines of the Office of Naval Research (ONR). Specifically this would amount to setting up directorships in major funding areas wherein the directors themselves have the authority to be proactive, i.e., to formulate and

exercise their own vision as to what research should be supported. Within this context, there should also be special funding set aside for high-risk ventures, with the understanding that the director is not to be penalized if a particular project fails to bear fruit. The performance of the director would be judged not only by his superiors but also by feedback actively solicited from the sector of academia in concern. Also as in ONR, NSF directorships should have indefinite duration, thereby allowing directors to lay plans and carry them out, and it should be possible for researchers to establish a relationship with the agency, based on past performance, rather than having to start anew every couple of years. It would also help to increase the overall budgets of these agencies; there are too many scientists whose talents are being wasted.

The latter aim constitutes a much more difficult issue, as the United States has strict rules regarding government involvement in private enterprise. It would seem, though, that some agency such as MITI could be established to aid corporations in identifying potentially fruitful lines of R&D and to grant incentives for working in those areas.

Whether projects such as LIFE are truly worthwhile has yet to be established, but more programs that promote collaboration between universities, government, and industry would certainly be in order. American ingenuity should be able to provide the needed new bridge between theory and applications.

Daniel G. Schwartz is an associate professor in the Department of Computer Science at Florida State University, Tallahassee. He holds BA and M.S. degrees in mathematics, and he completed his Ph.D. in systems science at Portland State University in 1981. His dissertation and subsequent research concerned formal axiomatizations of fuzzy logic and associated reasoning systems, whereas his more recent work has involved the development of a new symbolic approach to approximate reasoning. Dr. Schwartz has over 30 published articles on these and related topics. His current interests include the integration of fuzzy rule-based reasoning with other reasoning paradigms and their applications both in expert decision support systems and in automated control.

annealing, neural networks, etc., besides the traditional areas of control theory.

The other plenary lectures were given by Profs. B.D.O. Anderson of the Australian National University; Israel Gohberg of Tel Aviv University, Israel; and A. Isidori of the University of Roma, Italy.

Anderson's plenary talk was on optimizing the discretization of continuous time controllers. A standard problem in digital control is replacing a continuous time controller with a discrete time controller so that, as far as possible, closed-loop properties are preserved. The standard approaches to solve the problem do not make any use at all of the plant, whereas the closedloop properties clearly depend on the plant as well as the controller. Anderson proposed some novel approaches for solving the problem that make use of the knowledge of the plant. Two aspects of the closed-loop properties, stability and the closed-loop transfer matrix, are emphasized.

Gohberg's talk was on the interplay between interpolation problems and systems theory. He described some recent developments in interpolation theory and showed how these developments have greatly influenced modern theoretical systems theory research.

Isidori's talk was on robust regulation of nonlinear systems. Research in nonlinear systems is still in its infancy. Isidori showed how certain problems in nonlinear systems can be solved using linearization techniques while for others these techniques are not useful.

SPECIAL TOPICS LECTURES

The goal of the special topics lectures was to present research advances in certain selected areas of interest to the conference; the lectures were really designed for specialized subgroups of the participants.

The special topics lectures were as follows: