It has been reported that generalization performance of multilayer feedformard networks strongly depends on the attainment of saturated hidden outputs in response to the training set. Usually standard Backpropagation (BP) network mostly uses intermediate values of hidden units as the internal representation of the training patterns. In this letter, we propose construction of a 3-layer cascaded network in which two 2-layer networks are first trained independently by delta rule and then cascaded. After cascading, the intermediate layer can be viewed as hidden layer which is trained to attain preassigned saturated outputs in response to the training set. This network is particularly easier to construct for linearly separable training set, and can also be constructed for nonlinearly separable tasks by using higher order inputs at the input layer or by assigning proper codes at the intermediate layer which can be obtained from a trained Fahlman and Lebiere's network. Simulation results show that, at least, when the training set is linearly separable, use of the proposed cascaded network significantly enhances the generalization performance compared to BP network, and also maintains high generalization ability for nonlinearly separable training set. Performance of cascaded network depending on the preassigned codes at the intermediate layer is discussed and a suggestion about the preassigned coding is presented.

This paper describes the fundamental principle of novel push-push oscillators using hairpin-shaped split-ring resonators and their application to voltage controlled and injection locked oscillators for frequency synthesizers. The experimental results make it clear that the synthesizer systems discussed here have the advantages of high frequency operation, compact size and low power consumption. Experimental work has been carried out in the L band, but these systems can be applied to much higher frequencies.

Algorithms for computing channel capacity have been proposed by many researchers. Recently, one of the authors proposed an efficient algorithm using Newton's method. Since this algorithm has local quadratic convergence, it is advantageous when we want to obtain a numerical solution with high accuracy. In this letter, it is shown that this algorithm can be extended to the algorithm for computing the constrained capacity, i.e., the capacity of discrete memoryless channels with linear constraints. The global convergence of the extended algorithm is proved, and its effectiveness is verified by numerical examples.

When a new fabrication process is set up, especially in layout design for functional cells, of practical importance is how to make the best use of layout resources so far accumulated in old fabrication processes. Usually layout data of each element are expressed mostly in terms of positional coordinate values, and hence it is extremely tedious to modify them at every change of design rules for a new fabrication technology. To cope with this difficulty, the present paper describes an automatic recycling scheme for layout resources accumulated dedicatedly for functional cell generation. The main subject of this scheme is to transform given layout data into a layout description format expressed in layout parameters. Once layout data are parameterized, layout patterns of functional cells can be reconstructed simply by tuning up parameters in accordance with a new set of design rules. A part of implementation results are also shown.

A low-voltage, high-speed 4-bit CMOS single chip microprocessor, with instruction execution time of 1.0µs at a power supply voltage of 1.8V, has been developed. A single chip processor generally includes crystal oscillation circuits to generate a system clock or a time-base clock. But when the operating voltage is lowered, it becomes difficult to get oscillations to start reliably and to continue stably. This paper describes a low voltage circuit design method for built-in crystal oscillators. Simple design equations for oscillation starting voltage and oscillation starting time are introduced. Then effects of the circuit device parameters, such as power supply voltage, loop gain values, and subthreshold swing S, on the low voltage performance of the crystal oscillators are considered. It is shown that the crystal oscillators operate in a tailing (subthreshold) region at voltages lower than about 1.8 V. Subthreshold swing, threshold voltage, and open loop gain have a significant influence on low voltage oscillation capability. This design method can be applied to crystal oscillators for a wide range of operating voltages.

By adding a linear resistor in series with the inductor in Chua's circuit, we obtain a circuit whose state equation is topologically conjugate (i.e., equivalent) to a 21-parameter family C of continuous odd-symmetric piecewise-linear equations in R3. In particular, except for a subset of measure zero, every system or vector field belonging to the family C, can be mapped via an explicit non-singular linear transformation into this circuit, which is uniquely determined by 7 parameters. Since no circuit with less than 7 parameters has this property, this augmented circuit is called an unfolding of Chua's circuit--it is analogous to that of "unfolding a vector field" in a small neighborhood of a singular point. Our unfolding, however, is global since it applies to the entire state space R3. The significance of the unfolded Chua's Circuit is that the qualitative dynamics of every autonomous 3rd-order chaotic circuit, system, and differential equation, containing one odd-symmetric 3-segment piecewise-linear function can be mapped into this circuit, thereby making their separate analysis unnecessary. This immense power of unification reduces the investigation of the many heretofore unrelated publications on chaotic circuits and systems to the analysis of only one canonical circuit. This unified approach is illustrated by many examples selected from a zoo of more than 30 strange attractors extracted from the literature. In addition, a gallery of 18 strange attractors in full color is included to demonstrate the immensely rich and complex dynamics of this simplest among all chaotic circuits.

The mechanism of environment-dependent self-organization of "positional information" in a coupled nonlinear oscillator system is proposed as a new principle of realtime coordinative control in biological distributed system. By modeling the pattern formation in tactic response of Physarum plasmodium, it is shown that a global phase gradient pattern self-organized by mutual entrainment encodes not only the positional relationship between subsystems and the total system but also the relative relationship between internal state of the system and the environment.

We propose an integrated smart cable operation system and its architecture for the future cable network. In the proposed architecture, an application programs and various modules are loosely coupled using a cable operation system platform. We anticipate the task flows for the future optical cable network operation in order to realize the proposed system and architecture. Each task flow is broken down into "atomic tasks." The task flow can be changed easily by combining these atomic tasks. We use an object-oriented design for designing the cable operation system platform. As a first step towards the construction of the proposed system a pre-prototype system was constructed and the results are shown.

We report on Brillouin optical-fiber time domain reflectometry (BOTDR) for distributed temperature or strain measurement along a single-mode optical fiber. BOTDR uses Brillouin scattering in optical fibers, whose Brillouin frequency shift increases in proportion to temperature or strain induced in the fiber. This method requires access to only one end of a fiber, as with conventional optical time domain reflectometry (OTDR) which uses Rayleigh scattering in optical fibers. In BOTDR, a coherent optical detection method is used as a backscattered light detection technique. This technique can achieve both high sensitivity and high frequency resolution and easily separate a weak Brillouin line from a strong Rayleigh scattering peak and Fresnel reflected light. Experimental results show the potential for measuring temperature and strain distribution with respective accuracies of 3 or 0.006%, and a spatial resolution of 100m in an 11.57km long fiber.

Practical linewidth measurement accuracy better than 0.02 µm 3 sigma that meets the production requirement for devices with sub-half micron features, was achieved in a field emission scanning electron-beam metrology system (Hitachi S-7000). In order to establish high accuracy linewidth measurement, it was found in the study that reduction of electron-beam diameter and precise control of operating conditions are significantly effective. For the purpose of reducing electron-beam diameter, a novel electron optical system was adopted to minimize the chromatic aberration which defines electron-beam profile. As a result the electron beam diameter was reduced from 20 nm to 16 nm. In order to reduce measurement uncertainties associated with actual operating conditions, a field emission electron gun geometry and an objective lens current monitor were investigated. Then the measurement uncertainties due to operating conditions was reduced from 0.016 µm to 0.004 µm.

This paper deals with the theory and design method of an efficient radix-4 divider using carry-propagation-free adders based on redundant binary {-1,0,+1} representation. The usual method of normalizing the divisor in the range [1/2,1) eliminates the advantages of using a higher radix than two, bacause many digits of the partial remainder are required to select the quotient digits. In the radix-4 case, it is shown that it is possible to select the quotient digits to refer to only the four (in the usual normalizing method it is seven) most significant digits of the partial remainder, by scaling the divisor in the range [12/8,13/8). This leads to radix-4 dividers more effective than radix-2 ones. We use the hyperstring graph representation proposed in Ref.(18) for redundant binary adders.

This paper focuses on methodology underlying the application to fault tolerant computer systems with "no down communication" capability of stochastic Petri nets with general firing times. Based on a formal specification of the stochastic Petri net, we provide criteria for the marking process to be a regenerative process in continuous time with finite cycle-length moments. These results lead to strongly consistent point estimates and asymptotic confidence intervals for limiting system availability indices. We also show how the building blocks of stochastic Petri nets with general firing times facilitate the modeling of non-deterministic transition firing and illustrate the use of "interrupter input places" for graphical representation of transition interruptions.

The Hilbert curve is one of the simplest curves which pass through all points in a space. Many researchers have worked on this curve from the engineering point of view, such as for an expression of two-dimensional patterns, for data compression in an image or in color space, for pseudo color image displays, etc. A computation algorithm of this curve is usually based on a look-up table instead of a recursive algorithm. In such algorithm, a large memory is required for the path look-up table, and the memory size becomes proportional to the image size. In this paper, we present an implementation of a fast sequential algorithm that requires little memory for two and three dimensional Hilbert curves. Our method is based on some rules of quad-tree traversal in two dimensional space, and octtree traversal in three dimensional space. The two dimensional Hilbert curve is similar to the scanning of a DF (Depth First) expression, which is a quad-tree expression of an image. The important feature is that it scans continuously from one quadrant, which is obtained by quad tree splitting, to the next adjacent one in two dimensional space. From this point, if we consider run-lengths of black and white pixels during the scan, the run-lengths of the Hilbert scan tend to be longer than those of the raster scan and the DF expression scanning. We discuss the application to data compression using binary images and three dimensional data.

Periodic schedules are seldom treated in the theory but abound in practice (air flight schedule, train schedule, manufacturing schedule, etc). This paper introduces a Petri Net based perspective to periodic schedules. These are classified, according to the time interpretation into single-server and multiple-server semantics and, according to transitions firing periodicity constraints, into strict and general periodic schedules. Using a net transformation rule, the computation of the general schedule class can be done through techniques for the strict subclass. Introducing truncation error terms ε for the floor functions, a necessary and sufficient condition for the feasibility of a strict periodic schedule is given in terms of a large size system of nonlinear inequalities containing ε terms. Moreover averaging this condition on subperiods allows to get a small size linear system of inequalities as necessary conditions for speeding up iterative computation processes. This paper aims to present qualitative analysis of periodic schedules for deterministically timed Petri net systems, as a precursor to quantitative analysis that requires large-scale computational experiments and hence will be dealt in later work.

Gradient-based methods for the computation of the velocity from image sequences assume that the velocity field varies smoothly over image. This creates difficulties at regions where the image intensity changes abruptly such as the occluding contours or region boundaries. In this letter, we propose a method to overcome these difficulties by incorporating the information of discontinuities in image intensity into a standard local optimization method. The presented method is applied to the synthetic and real images. The results show that the velocity field computed by the proposed method is less blurred at region boundaries than that of the standard method.

We describe an approach to describe moving pictures in terms of their structural properties for video editing, video indexing, and video coding. The description contains 2D shape, motion, spatial relation, and relative depth of each region. To obtain the description, we develop the incremental segmentation scheme which includes dynamic occlusion analysis to determine relative depths of several objects. The scheme has been designed along the analysis-by-synthesis" approach, and uses a sequence of images to estimate object boundaries and motion information successively/incrementally. The scheme consists of three components: motion estimation, prediction with dynamic occlusion analysis, and update of the segmentation results. By combining the information from extended (longer) image sequences, and also by treating the segmentation and dynamic occlusion analysis simultaneously, the scheme attempts to improve successively over time the accuracy of the object boundary and motion estimation.

This paper describes a method for describing a three-dimensional (3-D) scene by integrating color and range data. Range data is obtained by a feature-based stereo method developed in our laboratory. A color image is segmented into uniform color regions. A plane is fitted to the range data inside a segmented region. Regions are classified into three types based on the range data. A certain types of regions are merged and the others remain unless the region type is modified. The region type is modified if the range data on a plane are selected by removing of the some range data. As a result, the scene is represented by planar surfaces with homogeneous colors. Experimental results for real scenes are shown.

The construction of an optical subscriber loop and its operation system will be one of the most important infrastructures for the information society of the future. This paper presents a discussion of current and future service trends, the evolution of fiber-optic systems in the loop, and subscriber loop operation systems. Several operation technologies are also discussed which will enable the operation systems described in this paper to be constructed. Of these, the key technologies are an operation system architecture with flexibility, software technologies based on object-oriented design and programming, and automatic operation modules.

In this paper, we discuss secure protocols for shared computation of algorithms associated with digital signature schemes based on discrete logarithms. Generic solutions to the problem of cooperatively computing arbitraty functions, though formally provable according to strict security notions, are inefficient in terms of communication--bits and rounds of interaction--; practical protocols for shared computation of particular functions, on the other hand, are often shown secure according to weaker notions of security. We propose efficient secure protocols to share the generation of keys and signatures in the digital signature schemes introduced by Schnorr (1989) and ElGamal (1985). The protocols are built on a protocol for non-interactive verifiable secret sharing (Feldman, 1987) and a novel construction for non-interactively multiplying secretly shared values. Together with the non-interactive protocols for shared generation of RSA signatures introduced by Desmedt and Frankel (1991), the results presented here show that practical signature schemes can be efficiently shared.

Mathematical morphology clarified geometrical properties of shape analysis algorithms for binary pictures. Results of labelling, distance transform, and adjacent numbering are, however, coded pictures. For full descriptions of shape analysis algorithms in the framework of mathematical morphology, it is necessary to extend morphological operations to code-labelled pictorial data. Nevertheless, extensions of morphology to code-labelled pictures have never discussed though the theory of gray morphology is well studied by several authors. Hence, this paper proposes a theory of the coded morphology which is based on the binary scaling of labels of pixels. The method uses n-layered binary sub-pictures for the processing of a picture with 2n labels. By introducing morphological operations for the coded point sets, we express some coding functions in the manner of the mathematical morphology. We also derive multidimensional array registers and gates which store and process coded pictures and morphological operations to them by proposing basic gates which compute parallelly logical operations for elements of Boolean layered arrays. These gates and registers are suitable for the implementation of the shape analysis processors on the three-dimensional VLSI and ULSI.