its application to talker recognition," Electr. and Comm. in Japan, vol. 57-A, pp. 5) and (b) last .... B.S. degree in physics from Brooklyn College,. NY, in 1935, and ...
IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS, VOL.
SMC-10, NO. 1, JANUARY 1980
(a)
t2
°Aug 2nd 1977
0 0
00
^ Jan 28h 1977
0
th
O Nov 15h 1976
A
o
o
-"""""l'~
0
o
_ t
(b)
57
1) detecting untypical training samples to be eliminated from further processing, 2) detecting natural subgroups to be accounted for as separate classes for increased linear classifier performance. In a simple example, the recognition rate for a linear classifier has been shown to increase from 80 to 96 percent when applying an automatic outlier elimination and subgroup detection. For multiterminal access control systems using speaker verification a small scale example has indicated that natural subgroups of common transfer function properties can be detected for all speakers involved in the experiment, allowing for a reduced similarity threshold for increased verification reliability.
Changes of the human vocal apparatus with time have been demonstrated by a cluster analysis of sample utterances produced in three sessions over a period of seven months. Further samples spoken with a head cold have been shown to form a distinct subgroup with extremely small variance, whereas samples spoken with closed nose seem to somehow bridge the gap between the "healthy" and the "head cold" cluster in feature space.
0
ACKNOWLEDGMENT The author appreciates the help of R. D. Naske and D. Voigt in providing the principal components analysis software.
0 0 0
00 0
0o
A a Mnv
'Id
1q77
cold heod cold"
_heod 0O
0 Aug.2"n d1977
.nose closed"
0 Aug
2nd l97
"healthy"
-
f
l______l____________ l_______________________ Fig. 6. Principal component analysis of (a) first three blocks (samples 1-30 in Fig. 5) and (b) last three blocks (samples 21-50 in Fig. 5).
[1]
REFERENCES Bunge, "AUROS-Automatic recognition of speakers by computers, principles of the speaker recognition system," presented at the 9th International E.
Congress on Acoustics, Madrid, Spain, 1977, paper 1 103. [2] H. H. Bock, Automatische Klassifikation. Gottingen: Vandenhoeck Rupprecht, 1974.
[3] J. H. Ward, "Hierarchical grouping to optimize an objective function," J. Am.
Stat. Ass., vol. 58, pp. 236-244, 1963. [4] D. Wishart, "An algorithm for hierarchical classifications," Biometrics, vol. 22, pp. 165-170, 1969.
[5] H. Spath, Cluster-Analyse-Algorithmen zur Objektklassifizierung und Datenreduktion. Munchen: Oldenbourg Verlag, 1975. [6] H. C. Andrews, Mathematical Techniques in Pattern Recognition. New York: Wiley, 1972. [7] S. Furui, "An analysis of long-term variation of feature parameters of speech and its application to talker recognition," Electr. and Comm. in Japan, vol. 57-A, pp. 34-42, 1974.
Book Reviews Management Applications of System Theory-C. V. Negoita (Basel: Birkhauser Verlag, 1979, 154 pp.). Reviewed by Dan Ralescu, Department of Mathematics, Indiana University, Bloomington, IN. 47405. a
At first sight, this book (vol. 57 in "Interdisciplinary Systems Research," series edited by S. Klaczko) seems to be a usual textbook devoted to
introducing the theory of discrete systems to students in management science. Looking more carefully at its content, one can see how new extensions are inserted on the backbone of the classical theory. Professor Negoita is the coauthor of a well-known monograph on fuzzy sets and systems [1] (vol. 11 in the same series). As I know, the main criticism of this well-received book was the lack of practical examples. One reviewer, H. Hamacher [2], in the first issue of the Journal of Fuzzy
Sets and Systems, notes that only a few exercises and examples of the real world applications are given and therefore the book cannot be considered as a textbook for students, but rather as an extraordinarily good reference book for the scientist. Although the title is not explicit, in the book under review Negoita seems to take into account these remarks in an attempt to illustrate the need for an algebraic theory of fuzzy sets. I will not dwell on the classical structure of this book (three chapters: state space approach, dynamics and behavior, optimal control), in order to comment more on the original contributions. The author says that many situations in real management can be explained by moven-ents of subjective evaluations. These movements can be described by a state equation, where the state is a fuzzy set. If we want to
58
IE
TRANSACTIONS ON SYSTeMS, MAN, AND CYRERNETICS, VOL. SMc-10, NO. 1, JANUARY 1980
represent the aggregation of partial evaluation, the state equation can be based on a min operator or on an addition operator. The author considers both these cases, introducing flexible and robust programming techniques. For an evaluation of fuzzy theories for decisionmaking, the reader is referred to [3] and [4]. In fact, the key point of the fuzzy approach is the lattice structure. The movement in the lattice is exploited to explain the confluence of constraints in fuzzy programming. The author uses the term pullback to describe a synthesis process and suggests that structural stability can be achieved by using partial evaluations. This is an excellent motive which can seduce many scientists to become engaged on the road of fuzzy systems. Structuralism, as method, can easily lead to formalizations. To understand all these subtleties, one must study all of the text, although in the author's paper presented at the 1978 Amsterdam Congress of Cyber-
netics and Systems [5] the reader could find an excellent abstract of the book. Finally, two words about the style. Clarity and plausibility take precedence over formal logic, although the author has not hesitated to be rigorous when important fine points are at stake. Everything considered, this is a really good book. REFERENCES [I] C. V. Negoii and D. A. Ralscu, Applicatiom of Fuzzy Sets to Systemu Analysis. Beu: Birkhiuser Verlag and New York: Halsted, 1975. [2] H. Hamacher, J. Fuzzy Sets and Syutenu, vol. 1, pp. 75,1978. (Book Review) (3] W. Kickert, Fuzzy Theories on Decsion Making. Leiden: M. NijhotZ 1978. [4] D. Raksu, 'A survey of the repreentation of fuzzy concepts and its applications, to apper in M. Gupta, R. Raade, and R. Yager, Eds.: Adeaces in Fuzzy Set Thery and Appi-caon.
[5]
Amsterdam: North-Holland 1979. C. V. Negotf. "On fuzzy systems," presnted at the 4th lt. Congress of Cyberneties and Systems, Amsterdam, 1978.
Contributors Alexander N. Christakis was the Head of the Futures and Policy Research Group at Battelle's Washington Operations. He is presently an Associate Professor in the Department of Engineering Systems and Science of the University of Virginia, Charlottesville. He is a policy planner with wide international experience in both the academic and professional fields. As a Fellow at the Academy for Contemporary Problems from 1972 to 1975, he was involved in 'futures research" and policy analysis on national and urban development issues, applying new methods for the exploration of the complexity of human settlements. He has also served as a consultant to the Club of Rome.
Lynn W. Ellis (S'48-M'49-SM'70-F'74) received the B.E.E. degree from Cornell University, Ithaca, NY, in 1948, the M.Sc. degree from Stevens Institute of Technology, Hoboken, NJ, in 1954, and the Doctor of Professional Studies in _ Management degree from Pace University, New York, NY, in 1978 with a dissertation on economies of scale in telecommunications. He is presently a Telecommunications Consultant. He was formerly Director of Research of International Telephone and Telegraph Corporation. In this position he supervised worldwide activities in applied research and advanced technology, and served as Executive Director of the ITT Research Council. Previously he was responsible for groups in that company's Headquarters Technical Department controlling worldwide development and engineering in wire and radio transmission, digital systems, telecommunications planning, and engineering standardization, following several years of narrower responsibility in that department. His 31 years of experience with ITT embraced technical work in the operating and manufacturing sectors of telecommunications as well as research and development in military electronics. Prior to his return to ITT Headquarters in 1966, he spent 11 years overseas in engineering management positions in ITT subsidiaries in Spain, England, and Australia. In this latter country he became Assistant Managing Director of ITrs principal Australia telecommunications manufacturing company, STC (Pty) Ltd. Dr. Ellis is a Fellow of the American Association for the Advancement of Science. He was a member of the Committee on Telecommunications ofthe National Academy of Engineering from 1973 to 1976, and Chairman of the Telecommunications Equipment Advisory Committee to the U.S. Department of Commerce from 1973 to 1975. He is currently a member of the IEEE Fellow Committee.
Samuel Globe (S'46-A'47-SM'55) was born in Brooklyn, New York, in 1916. He received the B.S. degree in physics from Brooklyn College, NY, in 1935, and the M.S. degree in Communication Engineering and the Ph.D. degree in Applied Physics from Harvard University, Cambridge, MA, in 1946 and 1959, respectively. From 1946 to 1955 he was an engineer and physicist at the Naval Ordnance Laboratory in Silver Spring, Maryland. From 1955 to 1956 he was a Senior Research Engineer with Chicago Midway Laboratories of the University of Chicago. From 1956 to 1962 he was Assistant for Research to the Vice-President of Avco's Advanced Development Division. Since 1962 he has been with the Columbus Laboratories of the BatteHle Memorial Institute in several positions. Since 1970 he has been Director of the Research Council. His research interests have included underwater acoustics, electronics, systems analysis, heat transfer, magnetohydrodynamics, and industrial innovation.
Kazaduiko Kawamwa (S'73-M'73)
was born in 1939 in Nagoya, Japan. He received the B.E. degree from Waseda University, Tokyo, Japan, in 1963, the M.S. degree from the University of California, Berkeley, in 1966, and the Ph.D. degree from the University of Michigan, Ann Arbor, in 1972, all in electrical engineering. From 1972 to 1973 he was a lecturer in the Department of Electrical Engineering, University of Michigan, Dearborn. Since 1973, he has been with Columbus Laboratories of the Battelle Memorial Institute, where he is a Principal Research Scientist within the Energy and Environmental Systems Assessment Section. His research interests have included the applcation of systems approach to societal problem solving, structural model building, technology assessment, longrange planning, R&D management, and futures research. In September 1979 he was awarded a Long-Term Senior Research Fellowship by Japan Society for the Promotion of Science. Dr. Kawamura is a member of the American Association For the Advancement of Science, the Society of Instrument and Control Engineers, and Sigma Xi. He is listed in American Men and Women of Science.
