Takao Kurokawa

Networks of Neural Nuclei

A new model of a neural system is presented. Neurons are considered to operate as parts of a neuronal ensemble called a “neural nucleus”, which has a homogeneous structure in connections among neurons. An excitatory nucleus shows one of the three fundamental properties in terms of the time course of the number of firing neurons in it: (i) increasing, (ii) constantly sustained and (iii) decreasing excitation. Mutually interacting structures of the neural nuclei are called nuclear networks. Simple nuclear networks consisting of one or a few nuclei are described to have functions such as integration, normalization and oscillation. In the nuclear networks inhibitory nuclei play an important role from the functional standpoint.

Analysis and classification of three-dimensional trunk shape of women by using the human body shape model

This paper proposes a new method for extracting shape components of the trunk of women from three dimensional measurements and tries a classification of the trunk shapes. Subjects in this study are 560 Japanese women, ranging in age from 19 to 63 years. First, the authors describe three-dimensional (3D) trunk shape using the control points given by fitting the human body shape model to 3D measurements of the subjects and reduce the number of the control points to be suited for statistical analysis based on correlation strength. The principal component analysis is applied to the shape data or the reduced set of the control points. Then, the authors interpret trunk shape components by combining factor loading map and averaged shape figures. Finally, the authors try to classify the trunk shape of Japanese women by means of the cluster analysis of component scores of the above results.

Adaptive Pattern Classifiers Applied to Time Optimal Control

Abstract This paper describes the adaptive methods to construct the optimal switching curves used for time optimal control of second order processes. Three types of adaptive pattern classifiers can be used, i.e.(1) gradient-variable piecewise linear, (2) zone-variable piecewise linear and (3) non-linear pattern classifiers. The adaptation logic introduced in this paper is based on the time course of geometrical relation between the state point and the switching curve in the state plane of the process and can be applied to the processes at work. The methods need neither a priori information of the process nor identifying it. Besides they are likewise applicable to both linear and nonlinear processes. It is shown that are arbitrary switching curve is adaptively optimalized, and that by the iterative application of this on-line adaptation, approximated optimal control can be realized.

In memory of Professor Hiroshi Tamura

Professor Hiroshi Tamura, who was honorary member of the Human Interface Society (Japan), passed away on the 28th of August 2010, at the age of 75. He was a visionary pioneer in the field of Human Interface in Japan and internationally. He was member of the Advisory Board of the Universal Access in the Information Society Journal. Professor Tamura was born in Niigata Prefecture on the 14th of April 1935. In 1958, he graduated in Electrical Engineering from Kyoto University. From the same university, he obtained his Master Degree in Engineering in 1960 and his PhD in Engineering in 1963. His academic and research career started in 1964 at the Faculty of Engineering of Kyoto University, but very soon he moved to the Faculty of Engineering Science of Osaka University, where he was appointed Assistant Professor in 1966. In 1988, he was appointed Professor at the Faculty of Engineering and Design of the Kyoto Institute of Technology, from where he retired as Honorary Professor in 1999. From 2001 to 2005, he was also Research Professor at the Faculty of Human Environment of the Hiroshima International University. Professor Tamura’s research interests were very wide. His PhD research was in the area of nonlinear oscillations. After moving to Osaka University, he extended his interests to oscillating and rhythmic phenomena in neurons and nerve systems, as well as nonlinear control by humans and animals. At the same time, he undertook research on man– machine systems and adaptive/learning control. These activities led to fundamental progress in the field, as he established the innovative result that humans’ nonlinear action behavior can be exploited for the control of complex systems, thus outperforming previous approaches based on humans as linear controllers. Subsequently, he conducted research in a wide variety of areas in the domain of bioengineering, including the nervous systems, animals’ learning, and exploratory eye movement. In the early 1980s, foreseeing the importance of Human Machine Interfaces in the years to come, he conceived the vision of establishing the field of Human Interface in Japan and internationally, and started a series of related activities. In 1983 and 1984, he organized two symposiums sponsored by the Kansai Chapter of the Society of Instrument and Control Engineers (SICE), entitled ‘‘Future Man– Machine Interface’’ and ‘‘Man–Machine Interface in Instrumentation and Control’’, respectively. Additionally, in 1984, he established the SICE ‘‘Technical Committee of Human Interface’’. The first Human Interface Symposium took place in 1985, and this symposium has been held annually since then. Although the term ‘‘Human Interface’’ was already in use in the United States of America at the time, Professor T. Kurokawa (&) Department of Systems Innovation, Graduate School of Engineering Science, Osaka University, Machikaneyama-cho 1-3, Toyonaka, Osaka 560-8531, Japan e-mail: kurokawa@sys.es.osaka-u.ac.jp

Description of human body shape using an isomorphic polygon

A novel polygon model describing human body shape is developed. Firstly, a method of expressing an arbitrary point on a polygon mesh using the UV plane is proposed. An isomorphic feature of the model enables us to fit the polygon model to three-dimensional (3D) measurements by means of the least square method. Secondly, in order to optimise the model structure, vertices of the model are allocated to trunk shape based on multiple sets of 3D measurements of Japanese women. Finally, the experimental results show that the derived generic model consisting of 1,087 vertices can reconstruct the original shapes with the average error of 1.19 mm, sufficient for practical use.

Simulation of brassiere-wearing figures

This article proposes a simulation method that combines a three-dimensional shape model of a human body with the genetic algorithm (GA) for estimating shape change of the breasts by wearing a brassiere. The model can describe trunk shape of any female with 750 control points for fitting the surface to many 3D points on the trunk surface. Because the model structure is common among women, their body shapes can be statistically analysed by using the control points. First, we related naked 230 breasts of 115 Japanese women with those wearing a full-cup brassiere through the multi-regression analysis. In this process, we searched the best regression formulas among a plethora of combinations of terms, or coordinates of the 49 control points on the naked models through the GA. Second, the regression formulas obtained above were applied to another 22 naked breasts to estimate their brassiere-wearing shape. Visual and numerical evaluation of the simulated shape revealed that this method could predict brassiere-wearing breast shape based on naked ones and could be expected to serve in a design process and for sales without the need to try on.

On reverberatory processes in homogeneous neuronal spaces

Reverberating neural activity is strictly defined and examined in continuous and discrete neuronal spaces with homogeneous structure. Reverberations start with a specific population of firing neurons called the initial excitation and spread out in waves of firing and refractory bands of neurons toward the periphery. The necessary and sufficient conditions for having reverberations are obtained for continuous space and discrete one-dimensional space. The excitation fronts of reverberating waves have stable shapes which depend only upon the structure of the neuronal space. The reverberatory processes in high-threshold discrete neuronal spaces show strongly nonlinear properties. Relation between reverberations and nervous functions is discussed.