In this paper, the generalized learning vector quantization (GLVQ) algorithm is applied to design a hand-written Chinese character recognition system. The system proposed herein consists of two modules, feature transformation and recognizer. The feature transformation module is designed to extract discriminative features to enhance the recognition performance. The initial feature transformation matrix is obtained by using Fisher's linear discriminant (FLD) function. A template matching with minimum distance criterion recognizer is used and each character is represented by one reference template. These reference templates and the elements of the feature transformation matrix are trained by using the generalized learning vector quantization algorithm. In the experiments, 540100 (5401 100) hand-written Chinese character samples are used to build the recognition system and the other 540100 (5401 100) samples are used to do the open test. A good performance of 92.18 % accuracy is achieved by proposed system.

Based on a previous work on handwritten Japanese kanji character recognition, a postprocessing system for handwritten Japanese address recognition is proposed. Basically, the recognition system is composed of CombNET-II, a general-purpose large-scale character recognizer and MMVA, a modified majority voting system. Beginning with a set of character candidates, produced by a character recognizer for each character that composes the input word and a lexicon, an interpretation to the input word is generated. MMVA is used in the postprocessing stage to select the interpretation that accumulates the highest score. In the case of more than one possible interpretation, the Conflict Analyzing System calls the character recognizer again to generate scores for each character that composes each interpretation to determine the final output word. The proposed word recognition system was tested with 2 sets of handwritten Japanese city names, and recognition rates higher than 99% were achieved, demonstrating the effectiveness of the method.

By adopting a network architecture in which not only a calling but also a called terminal can select a wavelength, a novel WDM network becomes possible. This we call Wavelength Assignment Photonic Network (WAPN). In this network wavelengths are a kind of network resources and according to requests from terminals, wavelengths are allocated or assigned to calls. In the system a wavelength used for a call is to be used for another call after the call is terminated. By supplying wavelengths to the home, a bitrate-free, protocol free or even transmission method free network can be realized. In this paper, from a viewpoint of S/N or Q factor, WAPN is evaluated with special focus on the node architecture--i. e. , from the viewpoint of node size, number of switching stages, crosstalk level,and losses, because the allowable node size is the crucial issue to decide the whole network capacity. After brief explanations of this proposed system, the model for system evaluations will be established and a node system is to be evaluated for some practical parameter values considering especially traffic characteristics of a node. As a result of this study a node system with capacity more than 100 thousands erl (about 20 Tbps throughput) can be constructed using present available technologies, which will enable us to construct large WAPN network with radius of 2,000 km and subscribers of about 50 millions.

By adopting a network architecture in which not only a calling but also a called terminal can select a wavelength, a novel WDM network becomes possible. This we call Wavelength Assignment Photonic Network (WAPN). In this network wavelengths are a kind of network resources and according to requests from terminals, wavelengths are allocated or assigned to calls. In the system a wavelength used for a call is to be used for another call after the call is terminated. By supplying wavelengths to the home, a bitrate-free, protocol free or even transmission method free network can be realized. In this paper, from a viewpoint of S/N or Q factor, WAPN is evaluated with special focus on the node architecture--i. e. , from the viewpoint of node size, number of switching stages, crosstalk level,and losses, because the allowable node size is the crucial issue to decide the whole network capacity. After brief explanations of this proposed system, the model for system evaluations will be established and a node system is to be evaluated for some practical parameter values considering especially traffic characteristics of a node. As a result of this study a node system with capacity more than 100 thousands erl (about 20 Tbps throughput) can be constructed using present available technologies, which will enable us to construct large WAPN network with radius of 2,000 km and subscribers of about 50 millions.

The problem of electromagnetic scattering caused by inductive discontinuities locate in parallel-plate waveguides, in particular when dealing with discontinuous conductors of finite thickness, is analyzed using the modified residue-calculus method, the equations suitable for a numerical calculation are derived. The incident wave is taken to be the dominant mode, and the reflection and transmission properties of a symmetrical inductive iris are discussed.

A method is proposed for deriving a traffic characteristics model that can be used to forecast the traffic volume for intelligent telecommunication services. A sort of regression analysis with dummy variables is used to represent the service quantitatively and to construct the traffic characteristics model. Recursive least squares estimation, which is a special case of the Kalman filter, is applied to the traffic characteristics model to forecast the traffic volume. In the proposed modeling and forecasting, qualitative factors representing a certain service attribute are selected and using an information criterion, the model with the best fit is identified as the most suitable forecasting model. Numerical results using practical observation data showed that the proposed method produces an accurate forecast and is thus effective for practical use.

Differences in the behavior of dispenser cathodes and oxide cathodes in laminar-flow type and crossover type electron guns were investigated by experiments and simulations under high-current-density conditions. When an oxide cathode is operated under such conditions, the heating effect due to Joule heat in the oxide layer exceeds the cooling effect, depending on the product of the work function and the cathode current, resulting in a rise in the cathode temperature. This rise in cathode temperature aggravates deterioration of emission characteristics during the life of an oxide cathode. In the case of the dispenser cathode, however, the cathode temperature decreases under high-current-density conditions. When an oxide cathode in a crossover type electron gun is operated, equipotential surfaces are formed in the curved surface in the oxide layer. The formation of an equipotential surface leads to relaxation of the loading. It is considered that this is the reason for the longer life of an oxide cathode in a crossover type electron gun than that of an oxide cathode in a laminar-flow type electron gun.

We fabricated a new lateral Power MOS FET with an SOI structure formed by a solid phase epitaxy method which has a buried gate under the channel layer and a trench gate/drain. We studied the temperature dependence of the Vg-Id characteristic and Vd-Id characteristic in the temperature range of from 27 to 150. In the Vd-Id characteristic, the drain current decreases as the temperature increases in the saturation region, but the drain current increases as the temperature increases in the linear region. Zero-temperature-coefficient bias point of Vg-Id characteristic was 0. 3 V. And, the threshold voltage variation was -2. 8 mV/. In the Vd-Id characteristic, the drain current decreases as the temperature increases. The specific on-resistance was obtained in the linear region of the Vd-Id characteristic. The specific on-resistance variation was 0. 3 mWmm2/. The temperature characteristics of this device are as good as those of the conventional MOS FET.

One-dimensional Cellular Automata (CA's) are considered as potential pseudorandom pattern generators to generate highly random parallel patterns with simple hardware configurations. A class of linear, binary, and of nearest neighbor (radius = 1) CA's is referred to here as elementary ones. This paper investigates operations of such CA's with fixed boundary conditions when non-null boundary values are applied to them. By modifying transition matrices of elementary CA's to include the influence of boundary values, structures of state transition diagrams are determined.

One of the important challenges in the design of ATM networks is how to provide quality-of-service (QoS) while maintaining high network resource utilization. In this paper, we discuss the role of real-time traffic characterization in QoS control for ATM networks and review several approaches to the problem of resource allocation. We then describe a particular framework for QoS control in which real-time measurements of a connection stream are used to determine appropriate parameters for usage parameter control (UPC). Connection admission control (CAC) is based on the characterization of the aggregate stream in terms of the individual stream UPC descriptors, together with real-time measurements.

A new type of fiber Fabry-Perot interferometer buried in a fiber connector housing was proposed. The transmission spectra revealed double peaks due to birefringence in the fiber and the peak separation showed a temperature dependence as large as -7. 7 MHz/deg, which was 2 orders of magnitude larger than that estimated from the thermal characteristics of its component materials.

Handprinted Chinese character recognition (HCCR) can be classified into two major approaches: statistical and structural. While neither of these two approaches can lead to a total and practical solution for HCCR, integrating them to take advantages of both seems to be a promising and obviously feasible approach. But, how to integrate them would be a big issue. In this paper, we propose an integrated HCCR system. The system starts from a statistical phase. This phase uses line-density-distribution-based features extracted after nonlinear normalization to guarantee that different writing variations of the same character have similar feature vectors. It removes accurately and efficiently the impossible candidates and results in a final candidate set. Then follows the structural phase, which inherits the line segments used in the statistical phase and extracts a set of stroke substructures as features. These features are used to discriminate the similar characters in the final candidate set and hence improve the recognition rate. Tested by using a large set of characters in a handprinted Chinese character database, the proposed HCCR system is robust and can achieve 96 percent accuracy for characters in the first 100 variations of the database.

A field theory for geometrical pattern identification is developed based on the postulate that various modified patterns are identified via invariant characteristics of pattern transformations. The invariant characteristics of geometrical patterns are written as the functional of the light intensity distribution of pattern, its spatial gradient, and also its spatial curvature. Some definite expressions of the invariant characteristic functional for two dimensional linear transformation are derived, and their invariant and feature extracting property are examined numerically. It is also shown that the invariant property is conserved even when patterns are deformed locally by introducing a "gauge field" as new degree of freedom in the functional in form of a covariant derivative. Based on this idea, we discuss a field theoretical model for pattern identification performed in biological systems.

This paper presents the results of the best differential characteristic search of FEAL. The search algorithm for the best differential characteristic (best linear expression) was already presented by Matsui, and improvements on this algorithm were presented by Moriai et al. We further improve the speed of the search algorithm. For example, the search time for the 7-round best differential characteristic of FEAL is reduced to about 10 minutes (Pentium/166 MHz), which is about 212. 6 times faster than Matsui's algorithm. Moreover, we determine all the best differential characteristics of FEAL for up to 32 rounds assuming all S-boxes are independent. As a result, we confirm that the N-round (7N32) best differential characteristic probability of FEAL is 2-2N, which was found by Biham. For N=6, we find 6-round differential characteristics with a greater probability, 2-11, than that previously discovered, 2-12.

This paper describes relation between the number of library pairs and error probability to have all the pairs as fixed points for second-order bidirectional associative memory (BAM). To estimate accurate error probability, three methods have been compared; (a) Gaussian approximation, (b) characteristic function method, and (c) Hermite Gaussian approximation (proposed by this paper). Comparison shows that Gaussian approximation is valid for the larger numbers of neurons in both two layers than 1000. While Hermite Gaussian approximation is applicable for the larger number of neurons than 30 when Hermite polynomials up to 8th are considered. Capacity of second-order BAM at the fixed error probability is estimated as the function of the number of neurons.

We propose an on-line writer verification method to improve the reliability of verifying a specific system user. Most of the recent research focus on signature verification especially in the field of on-line writer verification. However, signature verification has a serious problem in that it will accept forged handwriting. To overcome this problem, we have introduced a text-indicated writer verification method. In this method, a different text including ordinary characters is used on every occasion of verification. This text can be selected automatically by the verification system so as to reflect the specific writer's personal features. A specific writer is accepted only when the same text as indicated by the verification system is inputted, and the system can verify the writer's personal features from the inputted text. Moreover, the characters used in the verification process can be different from those in the enrolment process. This method makes it more difficult to get away with forged handwriting than the previous methods using only signatures. We also discuss the reliability of the proposed method with some simulation results using handwriting data. From these simulation results, it is clear that this method keeps high reliability without the use of signatures.

In this paper, we present CB-Power, a hierarchical power analysis and characterization environment of cell-based CMOS circuits. The environment includes two parts, a cell characterization system for timing, input capacitance as well as power and a cell-based power estimation system. The characterization system can characterize basic, complex and transmission gates. During the characterization, input slew rate, output loading, capacitive feedthrough effect and the logic state dependence of nodes in a cell are all taken into account. The characterization methodology separates the power consumption of a cell into three components, e.g., capacitive feedthrough power, short-circuit power and dynamic power. With the characterization data, a cell-based power estimator (CBPE) embedded in Verilog-XL is used for estimating the power consumption of the gates in a circuit. CBPE is also a hierarchical power estimator. Macrocells such as flip-flops and adders are partitioned into primitive gates during power estimation. Experimental results on a set of MCNC benchmark circuits show that the power estimation based on our power modeling and characterization provides within 6% error of SPICE simulation on average while the CPU time consumed is more than two orders of magnitude less.

This paper describes the new write/erase operation methods in order to improve the read disturb characteristics for Flash EEPROM cells which are written by channel hot electron injection and erased by F-N tunneling emission from the floating gate to the substrate. The new operation methods is either applying a reverse polarity pulse after each erase pulse or applying a series of shorter erase pulses instead of a long single erase pulse. It is confirmed that by using the above operation methods, the leakage current can be suppressed, and then the read disturb life time after 105 cycles write/erase operation is more than 10 times longer in comparison with the conventional method. This memory cell by using the proposed write/erase operation method has superior potential for application to 256 Mbit Flash memories as beyond.

Let L be any class of languages, L' be one of the classes of context-free, context-sensitive and recursively enumerable languages, and Σ be any alphabet. In this paper, we show that if the following statement (1) holds, then the statement (2) holds. (1) For any language L in L over Σ, there exist an alphabet Γ including Σ, a homomorphism h:Γ*Σ* defined by h(a)=a for aΣ and h(a)=λ (empty word) for aΓ-Σ, a Dyck language D over Γ, and a language L1 in L' over Γ such that L=h(DL1). (2) For any language L in L over Σ, there exist an alphabet of k pairs of matching parentheses Xk, Dyck reduction Red over Xk, and a language L2 in L' over ΣXk such that L=Red(L2)Σ*. We also give an application of this result.

This letter describes a new computational method to obtain the bifurcation parameter value of a limit cycle in nonlinear autonomous systems. The method can calculate a parameter value at which local bifurcations; tangent, period-doubling and Neimark-Sacker bifurcations are occurred by using properties of the characteristic equation for a fixed point of the Poincare mapping. Conventionally a period of the limit cycle is not used explicitly since the Poincare mapping needs only whether the orbit reaches a cross-section or not. In our method, the period is treated as an independent variable for Newton's method, so an accurate location of the fixed point, its period and the bifurcation parameter value can be calculated simultaneously. Although the number of variables increases, the Jacobian matrix becomes simple and the recurrence procedure converges rapidly compared with conventional methods.