|
1. |
A Composition Method to Realize Self‐checking Fault Locators |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 1-10
Yoshihiro Namiki,
Yoshiaki Koga,
Preview
|
PDF (672KB)
|
|
摘要:
AbstractA fault locator is a self‐checking functional unit which identifies a faulty module in a self‐checking system constructed with self‐checking modules (subsystems) by collecting and observing the test outputs from each module.Faulty modules are identified by observing the syndrome corresponding to the preset input code space and identifying to which module fault the input coding subspace corresponds. Since the inputs to the fault locator are given as vectors of the preset coding subspace in this case, there may not be a sufficient number of vectors to self‐test the fault locator. Thus we present a method to construct a self‐checking fault locator by using permuters for inputs to provide sufficient input vectors for sel
ISSN:0882-1666
DOI:10.1002/scj.4690190401
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
2. |
Reduction of Word and Minimal Phrase Candidates for Speech Recognition Based on Phoneme Recognition |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 11-22
Sho‐Ichi Matsunaga,
Masaki Kohda,
Preview
|
PDF (955KB)
|
|
摘要:
AbstractThis paper discusses the selection of candidates in speech recognition based on the phoneme recognition. The method is based on the result of phoneme recognition for the part of speech input, for which the segmentation is performed with a high reliability. Using the information concerning the order of the phonemes or phoneme chains, and the information concerning the top and tail phonemes, the candidates are selected. Since only the part for which the segmentation can be performed with a high reliability is used, the candidate reduction has a great effect for the clearly uttered speech, and vice versa. Consequently, the method has the feature that the recognition rate is degraded less by the candidate selection.First, the proposed selection method is introduced into the word recognition. The candidate selection is applied to all words in the dictionary. A recognition experiment was performed for the cases of the word dictionary composed of 643 city names, with 100 city names uttered by 50 examinees as the input. As a result, the word candidates were reduced to 16 percent, maintaining almost the same recognition performance as in the case without candidate reduction.Next, the proposed candidate selection is introduced into the phase recognition. In the method, the location of the phoneme to be rejected is estimated in the candidate selection in the derivation of hypothesis, and based on that result, the syntax tree is back‐tracked. An experiment was performed for 235 phrases uttered by 2 examinees. As a result, the phrase candidates were reduced to 21 percent, compared with the case without candidate selectio
ISSN:0882-1666
DOI:10.1002/scj.4690190402
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
3. |
Fault‐Tolerant Scheme on Partial Differential Equations with Nearest‐Neighbor Mesh‐Connected Machines |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 23-32
Gyo Osawa,
Toshio Kawai,
Hideo Aiso,
Preview
|
PDF (655KB)
|
|
摘要:
AbstractIn the numerical computation of a partial differential equation, it is useful to employ a parallel computer, where a large number of processors are connected in a mesh form. Such a scheme is now realizable with the advent of modern VLSI technology. On the other hand, there arises a request for the fault tolerancy with the increase of the processing units. When a partial differential equation is to be solved by such a computer, the requirement is not severe for the completeness of the data, since the iterative method usually is employed. This paper considers a problem of this nature, and discusses the recovery from the fault within the execution. The method is to detect the fault, to modify the connections among the processing units, separating the processing unit in fault, and to continue the computation by modifying a part of the discrete algorithm. By the proposed method, the dynamic recovery from the fault is easily realized, and the degradation of the computation accuracy due to the fault is suppressed. The proposed scheme is implemented on the parallel computer PAX‐32 with 32 processing units. The actual result of solving a two‐dimensional Poisson equation is presented, and a theoretical evaluation is made for the error in the one‐dimensional pr
ISSN:0882-1666
DOI:10.1002/scj.4690190403
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
4. |
An Experimental Evaluation of Classifiers Using Spatial Context for Multispectral Images |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 33-47
Takashi Watanabe,
Hitoshi Suzuki,
Preview
|
PDF (965KB)
|
|
摘要:
AbstractIn early classification of the multi‐spectral images obtained by remote sensing, pixelwise classification was used primarily. However, pixelwise classification is limited, noting only the spectral aspect of data. Several methods have been proposed to improve the classification performance by utilizing the spatial context, in addition to the spectral aspect. By contrast, few reports have been made which evaluate quantitatively those methods to present a systematic comparison. This paper considers the typical classifiers utilizing the local spatial context such as spatial continuity and spatial adjacency, which are compound decision theory, majority decision, adaptive classification and probabilistic relaxation, together with modified versions. An experiment was made to compare the classification accuracy of such classifiers. To make the quantitative evaluation objectively, artificial multispectral data simulating the LANDSAT image was used in the experiment. This paper presents the experimental result and its discussio
ISSN:0882-1666
DOI:10.1002/scj.4690190404
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
5. |
A Validation Method for Communication Protocols Using Context–Free Language Theory |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 48-55
Kaoru Kurosawa,
Yoshiki Kuwabara,
Takashi Shigeno,
Preview
|
PDF (413KB)
|
|
摘要:
AbstractUsually, communication protocols are modelled as a communication between two finite‐state machines. Yu and Gouda [1] have shown that if the messages of a single type and the finite‐state machine have no mixed nodes, the verification of deadlock can be made in a polynomial time. This paper shows that if the messages are of a single type, the verification of deadlock can be made in a polynomial time, even if there does exist mixed nodes. The proof is derived by indicating that the communication sequences arriving at deadlock states form a context‐free language. If there does not exist a deadlock in the protocol obtained by converting all the messages into a single type, no deadlock exists in the original protocol (the converse is not always true). In this sense, the result in this paper is impo
ISSN:0882-1666
DOI:10.1002/scj.4690190405
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
6. |
Optimization of Multiple Queries in Relational Database Systems |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 56-65
Jun'ichi Miyao,
Kazuyuki Tominaga,
Tohru Kikuno,
Noriyoshi Yoshida,
Preview
|
PDF (666KB)
|
|
摘要:
AbstractRecently, relational database systems which allow concurrent executions have been developed on mainframes and minicomputers. In such a system, optimization of multiple queries can be performed as an extension of optimization of a single query. By unifying the common operations in the multiple queries, the computation time or the number of disk page fetches can be reduced. In this paper, the optimization of the join operations is considered, since the join operation is the most time‐consuming one in the relational database. A problem (problem JM) is formulated to minimize the number of disk page fetches by unifying the common join operations in multiple queries. The number of page fetches depends on the realization method of a single join operation. In this paper, it is assumed that the join operation follows the nested loops method [4]. This paper proposes a heuristic algorithm for problem JM, and the effectiveness of the method is evaluated by a simulation experimen
ISSN:0882-1666
DOI:10.1002/scj.4690190406
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
7. |
Analysis on Memory Contentions in a Homogeneous Multiprocessor System |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 66-76
Yongjian Zhao,
Hiromi Okada,
Sadao Maekawa,
Preview
|
PDF (816KB)
|
|
摘要:
AbstractThe performance of a homogeneous multiprocessor system with common memory contentions is evaluated in this paper. A symmetric continuous‐time Markovian queueing network model is developed and analyzed. The main result is that the average number of active processors, as an important performance measure, is derived and given by a very simple recursive formula. Using the result, several properties of the multiprocessor system are proved analytically. It is found that a profound performance measure defined as processing power saturation ratio is significant in the evaluation of multiprocessor systems. Some numerical examples are also available to verify the obtained result
ISSN:0882-1666
DOI:10.1002/scj.4690190407
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
8. |
Phonemic Recognition of Input Frames by Partial Matching Method Based on Continuous Dynamic Programming |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 77-87
Ryu‐Ichi Oka,
Preview
|
PDF (760KB)
|
|
摘要:
AbstractThis paper proposes the partial matching method based on continuous DP, which is a new method to recognize the phoneme in continuous speech by pattern matching. The result of recognition experiment for the phoneme by the method is also shown. The standard pattern used in the continuous DP is a relatively macroscopic speech unit. The matching interval between the standard and the input patterns is labeled by a more microscopic speech unit, i.e., phoneme and pseudo‐phoneme. This is called the matching history, by the use of which, when the continuous DP values is less than a certain threshold, each input frame is identified using the phoneme and pseudo‐phoneme. For each input frame, the number of identifications for each phoneme or psuedo‐phoneme is calculated, and the result of recognition for the frame is determined as the phoneme or pseudo‐phoneme providing the maximum. The recognition experiment was made for the open data, composed of 19,440 frames uttered by 5 speakers, with 156 kinds of pseudo‐phonemes. The accumulated frame recognition rate was 57.6 percent for the first and 87.6 percent for the sixth. The accumulated pseudo‐phoneme recognition rate was 71.6 percent for the first and 90.1 percent for the sixth. The recognition system is of the frame‐wise form, and is suited to the realti
ISSN:0882-1666
DOI:10.1002/scj.4690190408
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
9. |
3–D Object Data Input System Using Rearview Mirrors |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 88-96
Takeshi Agui,
Tomohito Saito,
Masayuki Nakajima,
Preview
|
PDF (594KB)
|
|
摘要:
AbstractThis paper introduces a noncontact measurement technique for 3‐D data acquisition. The conventional optical projection method cannot measure a 3‐D object when a dead angle is encountered. This paper investigates a methods using rearview mirrors to increase the observability and decrease the dead angle of the projection point on the object. The rearview mirrors provide multiple view of the object in the same image, thereby simplifying the 3‐D data acquisition pr
ISSN:0882-1666
DOI:10.1002/scj.4690190409
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
10. |
Tongue‐Shaped Measurement Using Piezoelectric Thin Film |
|
Systems and Computers in Japan,
Volume 19,
Issue 4,
1988,
Page 97-105
Masafumi Matsumura,
Katsuhiko Fujii,
Preview
|
PDF (611KB)
|
|
摘要:
AbstractA tongue‐shaped measuring system was developed using a polymer piezoelectric film to analyze the tongue motion in harmonization of utterance. The copolymer piezoelectric film has a bimorph structure and produces a voltage proportional to the curvature of the tongue. By attaching the film on the tongue, the curvature at each point on the surface is measured, and the shape of the tongue is calculated from the curvature. To evaluate the shape measurement using the developed system, a simulator for the tongue shape was constructed, which has a curved surface of curvature between −0.4 to −0.57cm−1. The shape of the simulator was measured. It was verified as a result that the error of estimation along the normal direction of the surface is within 0.5 mm, which is sufficient for the measurement of the tongue shape. Then the developed system was applied to measure the shape of the median and the frontal shape of the tongue during utterance. It is shown that the shape of the tongue during utterance can be measured. In the case of median shape, the curving. process from the main part to the tongue tip during the utterance of /r/ was clearly measured. The frontal shape, which has been difficult to measure, can easily be measured as in the case of the median shape, by modifying the placement of the copolymer piezoelectric film. The concave shape on the median line of the tongue during vowel harmonization was observed
ISSN:0882-1666
DOI:10.1002/scj.4690190410
出版商:Wiley Subscription Services, Inc., A Wiley Company
年代:1988
数据来源: WILEY
|
|