|
1. |
Analysis and synthesis of continuous‐time hysteretic neural networks |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 1-8
Kenya Jin'no,
Toshimichi Saito,
Preview
|
PDF (428KB)
|
|
摘要:
AbstractIn this article the authors propose a continuous‐time neural network which includes bipolar hysteretic elements. They then consider the problem of synthesis in order to satisfy the following conditions: I) the system has no oscillating solutions; II) any desired patterns are stored as stable output vectors; and III) the total number of false stable output vectors is made as small as possible. The authors then prove a sufficient condition for I) and the necessary and sufficient conditions for II). Based on these results, they propose a synthesis procedure which is also effective for III). The domain of attraction of stable outputs may be easily calculated using a fast algorithm using exact solutions. The efficiency of the procedure is confirmed through experimental result
ISSN:1042-0967
DOI:10.1002/ecjc.4430760401
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
2. |
A method of estimating the irregularity of scattering medium using characteristics of ultrasonic speckle |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 9-16
Akihisa Ohya,
Junji Kashioka,
Masato Nakajima,
Preview
|
PDF (625KB)
|
|
摘要:
AbstractWhen examining a random group of scatterers, random granular patterns called speckle appear on echograms obtained by pulse echo ultrasonic diagnostic equipment. In living body organs such as the liver, there is a profound connection between the irregularity of the tissue structure and the severity of the disease. Therefore, if the degree of the irregularity can be estimated through analysis of the echogram, the possibility exists that the properties of the tissues can be diagnosed quantitatively.This paper considers a medium that grouped scatterers as the model of the object and investigates the relationship between the statistics of interference pattern such as the speckle that appears on the ultrasonic echogram and the irregularity of the scattering medium. Also examined is the possibility of estimating the degree of irregularity of the medium from the echogram.From the results of a computer simulation, it is found that if the density of the scatterers is known, the degree of irregularity of the medium can be determined from the value of the signal‐to‐noise ratio (SNR) statistic of the echogram. In addition, it is learned that by using the SNR as the statistics together with the skewness of the brightness histogram of the image, it is possible to estimate the degree of irregularity and the density of the scatter
ISSN:1042-0967
DOI:10.1002/ecjc.4430760402
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
3. |
Learning rules of neural networks for inverse systems |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 17-23
Yutaka Maeda,
Preview
|
PDF (495KB)
|
|
摘要:
AbstractThis paper considers on‐line learning of neural networks for an inverse problem. A neural network must learn an inverse system of a static unknown system on the basis of an evaluation function that is a squared error between an output of the unknown system and its desired value.Usual learning methods need information about the sign of the sensitivity of the unknown system. This paper proposes learning rules that do not require the information about that sign. These rules utilize a perturbation signal. Rules with the perturbation on an input of the unknown system and a rule with it on each weight of the neural network are presented. The usefulness of these learning rules is confirmed via some numerical simulation
ISSN:1042-0967
DOI:10.1002/ecjc.4430760403
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
4. |
Satellite constellations for a multiple LEO satellite network |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 24-34
Tomonori Usui,
Tsutomu Kawabata,
Yoshikuni Onozato,
Ikuo Oka,
Preview
|
PDF (749KB)
|
|
摘要:
AbstractRecently, research has been very active in communication networks that could service the entire globe by using multiple, low, earth‐orbiting satellites. This has been spurred on by progress in the field of digital communication technique, technical advances of satellite launching, and a decrease in the costs of small‐sized satellites.In this paper, the construction of low, earth‐orbiting satellite communication networks is discussed from the viewpoint of the design of orbits and satellite positions. First, a symmetric orbit model derived from the idea of regular dodecahedron is proposed. Next, two evaluation methods of satellite distribution are shown. One is the method of summing the squares of distances between satellites and their earth stations. This is a general evaluation method based on the satellite orbits and satellite positions. The other is an evaluation method of the coverage based on the altitude of satellites and the antenna angles of the earth station. This method is applied to the proposed symmetric orbit, and characteristics of different satellite deployments are discussed comparat
ISSN:1042-0967
DOI:10.1002/ecjc.4430760404
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
5. |
Analysis of chaotic neuron models with information theory |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 35-46
Kunihiko Ishiyama,
Susumu Itoh,
Toshio Utsunomiya,
Kazuyuki Aihara,
Preview
|
PDF (814KB)
|
|
摘要:
AbstractA one‐dimensional mapping of a chaotic neuron model which generates deterministic chaos is analyzed from the viewpoint of the information theory.First, the mutual information between the initial value and the value after iterations of the mappings is calculated, and it is shown that the information of the initial value decays almost exponentially. To examine the decaying behavior in detail, the dynamics of the chaotic neuron model is represented as the information flow of “0” and “1” in a bitwise space (register) and the behavior of the mutual information contained in each bit is examined. It is shown through this analysis that the information in the chaotic neuron model is maintained, being mixed and distributed among the bits. Moreover, the chaotic neuron model is compared with two typical chaotic dynamical systems with different information structures, i.e., the B‐Z mapping and the logistic mapping to clarify the information structure. Finally, it is shown for the system, where the chaotic neuron models are unidirectionally connected as a one‐dimensional network, that the information transmission is made possible by its peculiar informat
ISSN:1042-0967
DOI:10.1002/ecjc.4430760405
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
6. |
Testable design for cell‐based design ASICs and its application to VLSI image signal processor |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 47-56
Hiroyuki Kawai,
Shin‐Ichi Nakagawa,
Masahiko Yoshimoto,
Yasutaka Horiba,
Takuji Ogihara,
Akiharu Tada,
Preview
|
PDF (859KB)
|
|
摘要:
AbstractThis paper proposes a testing method which provides an efficient test for the library macrocell built in a large‐scale and high‐speed application‐specific IC (ASIC) of a cell‐based design, especially in the parallel processor for the image processing VLSI. The proposed method consists of the test pattern assembly tool and the testability design for the efficient test. The test pattern is assembled automatically from the test pattern library of the macrocell and the access sequence description, which describes the access procedure from the external pins of the LSI and/or scan registers to the input/output pins of the macrocell built in the LSI.By the foregoing description, the access to the macrocell through the sequential circuit also is possible. With regard to the design for testability, the clock control method, the bus driver circuit, and the test mode control method are newly devised and are combined with the system bus and the scan register.The proposed method is applied to the high‐speed image processing VLSI. The test patterns for four library macrocells in the LSI (two RAMs, multiplier, and PLA) were assembled in one day. The proposed method can also be applied to the user developed macrocell which is utilized iteratively in an LSI. The increase of the number of transistors and the transistor area is suppressed to 3.1 percent (in regard to logic unit) and 0.6 percent (in regard to the whole chip area), respectively. It is verified that there is little decrease of the operational speed in the system operation and does not affect the high‐speed image
ISSN:1042-0967
DOI:10.1002/ecjc.4430760406
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
7. |
A parallel automatic routing technique based on local current comparison method—application of associative silicon retina |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 57-70
Gui‐Xin Cheng,
Mamoru Tanaka,
Preview
|
PDF (973KB)
|
|
摘要:
AbstractThis paper proposes a new local current comparison method to search for the routing path. In the proposed method, the routing plane is represented by a resistive network. A voltage is provided to the start and the end pins to be connected, and the routing path is determined from the current distribution in the equilibrium state. Thus, the local current of the edges connected to each node of the routing path is maximized. By this local current comparison method, the routing path is determined along the direction of the maximum current and several nets can be routed while avoiding the competition.The routing method for the multipin net also is proposed based on the local current comparison method. Since the potential distribution of the nodes in the resistive network is an approximate solution of the equipotential line distribution of the static electric field, it is shown that the optimal or quasi‐optimal short routing path can be determined. Using the multilayer resistive network, the multilayer routing can also be realized. Then the automatic routing method is proposed where more than one net can be routed in parallel by modifying the specification of voltages according to the result of the previous current distribution obtained by the local current comparison method.A method to locate the bottleneck (BN) of the net also is presented. The proposed local current comparison method is related closely to the node neighborhood analysis from the viewpoint of the cellular neural net, based on the energy minimization method. The method is especially suited to the dedicated hardware implementation by the mixed analog‐digital c
ISSN:1042-0967
DOI:10.1002/ecjc.4430760407
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
8. |
Reliability of maximum‐likelihood decoded codewords and its application to decoding of concatenated codes |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 71-78
Hatsukazu Tanaka,
Preview
|
PDF (543KB)
|
|
摘要:
AbstractThis paper attempts to improve the reliability of the maximum‐likelihood decoded codeword. The reliability function is defined anew based on the Euclid distance in the signal space and is derived for the case where the signal representation for the binary linear code is precisely specified. The traditionally well‐known Viterbi algorithm is modified to evaluate the reliability based on the proposed reliability function in parallel with the maximum‐likelihood decoding. This algorithm is called the modified Viterbi algorithm.To verify the usefulness of the newly defined reliability, the difference between the reliability and the likelihood is examined. The block‐error rate when the algorithm is applied to the decoding of the concatenated code is compared to the block‐error rates of the various conventional decoding methods. As a result, the block‐error rate is improved drastically by the decoding which effectively utilizes the reliability function; and it has been shown to be useful to introduce the concept of reliability into the decod
ISSN:1042-0967
DOI:10.1002/ecjc.4430760408
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
9. |
A generalized principal component analysis for mixed measurement level data |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 79-91
Hiroyuki Yaguchi,
Manabu Ichino,
Preview
|
PDF (828KB)
|
|
摘要:
AbstractThe principal component analysis is well known as a method to summarize the phenomenon described by the quantitative features. The quantification theory III is known as a method to summarize the phenomenon described by the mixed qualitative and quantitative features. However, several problems are pointed out in the quantification theory III, e.g., the manipulation containing an arbitrariness to divide the quantitative feature into intervals or the increase of the number of features due to the use of the dummy variables.This paper proposes a generalization of the principal component analysis for the mixed qualitative and quantitative data. A model is considered which can handle the mixed qualitative and quantitative data, and the quantification of the information is executed on the model using the distance function. Then the principal component analysis is applied. The proposed method is a natural extension of the ordinary principal component analysis and has the advantages that the resolution for the score is high, the number of features is less, and the interpretation of the result of analysis is easy. Using two simple examples, the proposed method is compared to the quantification theory III and the effectiveness of the proposed method is demonstrated.
ISSN:1042-0967
DOI:10.1002/ecjc.4430760409
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
10. |
A method of selecting HDTV test pictures for subjective quality assessments |
|
Electronics and Communications in Japan (Part III: Fundamental Electronic Science),
Volume 76,
Issue 4,
1993,
Page 92-105
Takeshi Tachi,
Satoru Iai,
Nobuhiko Kitawaki,
Preview
|
PDF (1199KB)
|
|
摘要:
AbstractThis paper presents a quantitative selection method for still images for use in assessment tests to obtain reasonable assessment results for high‐quality images in subjective assessment tests of high‐definition television (HDTV) signals. First, based on the selection criterion which is to select images that exhibit the characteristic of high quality, it was clarified that the three elementary psychological factors of vividness, comfortableness, and sharpness constituted the characteristics of high‐quality images and that the amplitudes of these psychological factors could be quantified by means of the statistics of the HVC signal distribution in the Munsell color space of the image.Next, based on the selection criterion where images can be evaluated accurately even when slightly degraded, subjective assessment tests were performed for random additive noise, high‐frequency cutoff, and adaptive DCT coding distortion. The results showed clearly that images suited for use in the assessment test of each degradation could be selected based on the statistics of the image HVC signal distribution. By performing assessment tests using the smallest possible number of images selected by the guidelines presented here, assessment results can easily be obtained efficiently and r
ISSN:1042-0967
DOI:10.1002/ecjc.4430760410
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1993
数据来源: WILEY
|
|