The performances of BPNN (neural network trained by back propagation) and PCANN (neural network which computes principal component analysis) for ECG data compression have been investigated from several points of view. We have compared them with an existing data compression method TOMEK. We used MIT/BIH arrhythmia database as ECG data. Both BPNN and PCANN showed better results than TOMEK. They showed 1.1 to 1.4 times higher compression than TOMEK to achieve the same accuracy of reproduction (13.0% of PRD and 99.0% of CC). While PCANN showed better learning ability than BPNN in simple learning task, BPNN was a little better than PCANN regarding compression rates. Observing the reproduced waveforms, BPNN and PCANN had almost the same performance, and they were superior to TOMEK. The following characteristics were obtained from the experiments. Since PCANN is sensitive to the learning rate, we had to precisely control the learning rate while the learning is in progress. We also found the tendency that PCANN needs larger amount of iteration in learning than BPNN for getting the same performance. PCANN showed better learning ability than BPNN, however, the total learning cost were almost the same between BPNN and PCANN due to the large amount of iteration. We analyzed the connection weight patterns. Since PCANN has a clear mathematical background, its behavior can be explained theoretically. BPNN sometimes generated the connection weights which were similar to the principal components. We supposed that BPNN may occasionally generate those patterns, and performs well while doing that. Finally we concluded as follows. Although the difference of the performances is smal, it was always observed and PCANN never exceeded BPNN. When the ease of analysis or the relation to mathematics is important, PCANN is suitable. It will be useful for the study of the recorded data such as statistics.

This paper presents an equation capable of briefly evaluating the length of white noise sequence to be sent as a training signal. The equation is formulated by utilizing the formula describing the convergence property, which has been derived from the IIR filter expression of the NLMS algorithm. The result revealed that the length is directly proportional to I/[K(2-K)] where K is a step gain and I is the number of the adaptive filter taps.

In this paper, we propose a polynomial time algorithm for computing the expected maximum number of vertex-disjoint s-t paths in a probabilistic basically series-parallel directed graph and a probabilistic series-parallel undirected graph with distinguished source s and sink t(st), where each edge has a mutually independent failure probability and each vertex is assumed to be failure-free.

Fuzzy control is suitable for automotive control, because fuzzy control achieves controllability as good as control by humankind. However, since automotive control requires milli-second response and learning control, and the fuzzy system in automobiles requires fewer components (built-in type), a custom fuzzy inference LSI is needed for automotive control. We then indicated requirements of a fuzzy inference LSI suitable for automotive control and fabricated a fuzzy inference LSI using 1.5 µm CMOS process technique. This fabricated fuzzy LSI is designed to utilize in various automotive control experiments such as engine control, cruise control, brake control and steering control. The number of input variables is six, the number of output variables is two, the maximum number of production rules is 256, and the inference time is 63 microseconds (under the condition of six inputs, two outputs and 256 rules). The features of the fuzzy LSI are high speed inference, a built-in type, learning control ability and a memory structure separating into a rule memory and a membership function memory. A fuzzy control system is implemented only by the addition of two devices: the fuzzy LSI and an EPROM. The fuzzy LSI was applied to a rough road durability test aiming at the automatic driving equivalent to the human driver operation. In the test, fuzzy control and linear control were compared in terms of the compensation steering degrees. Linear steering control had a high rate of compensation steering of less than thirty degrees. On the other hand, the accumulated steering compensation of less than twenty degrees in the fuzzy control was about one third that in the linear control. The fuzzy steering control had the same steering compensations as that of human steering. The fuzzy LSI fabricated for various experiments is too large (10.7 mm10.9 mm) to adopt as automotive parts. Therefore, we studied a smaller-sized fuzzy LSI by limiting functions, by changing the parallel processing into sequential processing and by thinning out the memory data of input membership functions. The number of input variables is four, the number of output variables is two, the maximum number of production rules is 160 and the expected inference time is 140 micro-seconds (in the worst case). The obtained chip is small enough (4.8 mm4.8 mm) for automotive applications. Since the chip contains all the memories that are needed to execute fuzzy inference, the chip can be built in a microprocessor as a fuzzy inference co-processor without any other circuits.

In a previous article, an optical array imaging system has been presented. In this system, first, a set of array data is collected by repeatedly illuminating the object with laser light from each array element, detecting the reflected light as interferogram, and extracting the reflected wave field based on the spatial heterodyne detection. Then, an eigenvalue analysis is applied to the data to derive the wave field that would backpropagate and focus at a single point on the object; in this case, the iterative algorithm is used which indicates that the object point may have the largest reflectivity. It was shown experimentally that the single-point-focusing was attained for objects having several such parts with almost the same reflectivities. A preliminary study by computer simulation, however, indicates that the probability with which the wave focuses at multiple object points would not be small enough, resulting in a degraded image with ghost image components. In this paper, the array data within subaperture regions are selectively used to attain the single-point-focusing and obtain a good image for any object. First, it is shown analytically that the change in the dimension or center position of the aperture is effective to change the eigenvector so that it attains the single-point-focusing. Then, a procedure to find the optimum subapertures and a measure evaluating the degree of single-point-focusing for the eigenvector are presented. The method is examined in detail using experimentally obtained array data, and the results show that the method is effective in obtaining good images for any objects without sacrificing image resolution. When we compare the imaging system to an automatic focusing camera, it may be said that the additional processings enhance the capability of automatic focusing to a great degree.

We have developed an SIMD processor on a double-height VME board. We achieved a good balance between cost and performance by combining four identical gate-array LSIs in the processor array with a 16-bit degital signal processor (DSP), standard dynamic random-access memories (DRAMs) and other peripherals. The gate-array LSIs have 168-bit processing elements (PEs), each containing a one-bit processing block and a serial multiplier. This PE structure offers high-level bit processing capability and peak performance of 512 million operations per second (MOPS) for 8-bit multiply and accumulate operations. Effective performance of more than 300 MOPS for 8-bit array data processing is achieved by using an LSI structure tuned to the DRAM access rate, although the processing speed is reduced by the DRAM access bottleneck. The LSIs also have two unique additional hardware structures that speed up various array data processes. One is an inter-PE routing register array for supporting a transmission, rotation and memory access path. The other is a tree-structure network for propagating operations among PEs. With these cost-effective structures, the SIMD processor is expected to be widely used for two-dimensional data processing, such as image processing and pattern recognition.

A technique was developed to reduce ECG data efficiently within a controlled accuracy. The sampled and digitized data of the original waveform of an ECG is transformed in three major processes. They are the calculation of a beat-to-beat variation, a polygonal approximation and the calculation of the difference between consecutive node points. Then, an adaptive coding technique is applied to minimize redundancies in the data. It was demonstrated that the ECG waveform sampled in 200 Hz, 10 bit/sample, 5 µV/digit could be reduced with the bit reduction ratio of about 10% and within the reconstruction error of about 2.5%. A polygonal approximation method, called MSAPA, was newly developed as a modification of the well known method, SAPA. It was shown that the MSAPA gave better reduction efficiency and smaller reconstruction error than the SAPA, when it was applied to the beat-to-beat variation waveform. The importance of the low-pass filtering as a preprocessing for the polygonal approximation was confirmed in concrete examples. The efficiency of the proposed technique was compared with the cased in which the polygonal approximation was not used. Through these analyses, it was found that the redundancy elimination of the coding technique worked effectively in the proposed technique.

The two-dimensional nested cellular automata array presented here as a method for testing CCD arrays accommodates a set of spatial bilateral inhibition and excitation, and thus generates spatio-temporal artificial chaotic signals. Adequate use of the spatio-temporal pulses achieves exact line detection that is completely different from the template-matching scheme used by conventional methods.

We extend the concept "optical instanton" to arbitrary dielectric media. For these general cases the exact analytical approach is no longer available. We derive an approximate analytical solution that would be valid in the vicinity of the light cone. A comparison is made between the analytical and the numerical solutions.

In this article, the autocorrelation associative neural network that is one of well-known applications of neural networks is improved to extend its capacity and error correcting ability. Our approach of the improvement is based on the consideration that negative self-feedbacks remove spurious states. Therefore, we propose a method to determine the self-feedbacks as small as possible within the range that all stored patterns are stable. A state transition rule that enables to escape oscillation is also presented because the method has a possibility of falling into oscillation. The efficiency of the method is confirmed by means of some computer simulations.

In the present paper, we focus ourselves on the turning point (TP) algorithm proposed by Mueller and evaluate its performance when applied to a Gaussian signal with definite covariance function. Then the ECG wave is modeled by Gaussian signals: namely, the ECG is divided into two segments, the baseline segment and the QRS segment. The baseline segment is modeled by a Gaussian signal with butterworth spectrum and the QRS one by a narrow-band Gaussian signal. Performance of the TP algorithm is evaluated and compared when it is applied to a real ECG signal and its Gaussian model. The compression rate (CR) and the normalized mean square error (NMSE) are used as measures of performance. These measures show good coincidence with each other when applied to Gaussian signals with the mentioned spectra. Our results suggest that performance evaluation of the compression algorithms based on the stochastic-process model of ECG waves may be effective.

The Traveling Salesman Problem (TSP) can be solved by a neural network using the coding scheme based on the adjacency of city in the tour. Using this coding scheme, the neural network generates a better solution than that using other coding schemes. We, however, often get the invalid solution consisting of some subtours. In this article, we propose a method of eliminating subtours using additional neurons. On the computer simulation it is shown that we get the optimum solution by means of taking only O(n2) additional neurons and trials.

Electronics and automobiles were bound together by the introduction of emission regulations in the 1970's. The rapid progress of control technology and semiconductors that typify microcomputers has brought still closer relations between them. Without electronics, it would be impossible to realize features such as pursuit of comfort and environmental and safety measures which should be added to the automobile's fundamental features. In looking ahead to the future, the role of electronics in achieving electric automobiles and the ultimate goal of "automatic driving" is ever-increasing. Everyone knows that automobiles have become indispensable in our lives. In the future, the role of electronics will become increasingly important in order to evolve automobiles even further to allow harmonization with society.

This letter shows that VEPs can be easily measured by using color cards as the color stimulus, and that the responses evoked by a difference in chroma could be described largely by the value of the first principal component in principal component analysis.

For the effective control of microwaves in the frequency domain, we propose a new method utilizing current distributions of standing waves on the terminated microstrip line. We analized a short ended microstrip line using the (FD)2TD method to indicate the effectiveness of our proposal. Further we proposed an optically controlled microstrip filter as an application of this method.

This paper gives a systematic approach to generate a Markov chain by a discrete-valued auto-regressive equation, which is a a nonlinear auto-regressive equation having a discrete-valued solution. The power spectrum, the correlation function and the transition probability are explicitly obtained in terms of the discrete-valued auto-regressive equation. Some computer results are illustrated in figures.

We analyze the nonlinear dynamics of PLL from the "complex" singularity structure by introducing the complex time. The most important results which we have obtained in this work are as follow: (1) From the psi-series expansion of the solution, the local behavior in the neighbourhood of a movable singularity is mapped onto an integrable differential equation: the Ricatti equation. (2) From the movable pole of the Ricatti equation, a set of infinitly clustered singularities about a movable singularity is shown to exist for the equation of PLL by the multivalued mapping. The above results are interesting because the clustering and/or the fractal distribution of singularities is known to be a characteristic feature of the non-integrability or chaos. By using the method in this letter, we can present a circumstantial evidence for chaotic dynamics without assuming any small parameters in the equation of PLL.

An omnidirectional microstrip antenna using a parasitic cylinder is presented. A rectangular patch is formed on a dielectric substrate and it's completely covered with an aluminum cylinder which is somewhat shorter than a half of free space wavelength. Under such configuration the aluminum cylinder works as a parasitic element. This antenna can provides uniform omnidirectional radiation patterns and a broad frequency bandwidth. In this paper an experimental method for designing such an element is described. Measured input impedance characteristics, current distribution around the surface of the cylinder and patterns are also shown. By properly adjusting the coupling intensity between the patch and the parasitic cylinder a broad bandwidth antenna element can be realized. Some methods to adjust the coupling intensity are shown. A wide bandwidth element up to 14% for VSWR1.5 is obtained. Arranging many patches lengthways on a substrate and placing metallic cylinders around each patches, we can realize a high-gain and broad bandwidth collinear antenna.

This letter proposes a microstrip band elimination filter having an optically controlled small gap on a resonant section for the shift of the eliminated frequency range using the semiconductor plasma. The basic characteristics of this filter are analized theoretically utilizing the (FD)2TD method.

This paper describes and analyzed several indices in assessing algorithms of data compression of electrocardiograms, such as the cross correlation (CC), the percent root mean square difference (PRD), and a new measure of standardized root mean square difference (SRD). Although these indices are helpful to objectively evaluate the algorithms, the visual examination of the reconstructed waveform is indispensable to decide the optimal compression ratio. This paper presents the clinical significance of selected waveforms which are prone to be distorted or neglected in the restored waveforms but are crucial for cardiologists to diagnose the patient. A database of electrocardiograms is also proposed for the comparative evaluation of compression algorithms.