摘要:

AbstractWe have been studying the parallel coordinated computing system composed of several AI workstations ELIS, as well as the knowledge processing on that system. From the viewpoint of the modularity and the hierarchy, an object‐oriented approach is adopted as the fundamental paradigm. In the knowledge representation on such a system, the object for representing the knowledge must be modified in a dynamical way since the knowledge always changes by itself.In this paper, such a modifiable object is called a dynamic object. In the traditional system, however, the object has been considered as static, and instance variables, for example, cannot be added or deleted dynamically during execution. The dynamic object is realized by modifying the object mechanism of TAO, which is the kernel language of ELIS.This paper proposes a dynamic object‐oriented programming based on this dynamic object, which is suited to the representation of the changing knowledge. It is indicated that from the standpoint of the dynamic knowledge representation, the dynamic object‐oriented programming is more natural and more efficient than the traditional object‐oriented prog

ISSN:0882-1666    

DOI:10.1002/scj.4690210501

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractThis paper evaluates the EM‐3, a data‐driven machine with advanced control mechanism. The advanced control mechanism with pseudo‐result concept has been introduced into the data‐driven mechanism. To evaluate this mechanism on a parallel processing environment, the prototype machine has been developed. Five benchmark programs, which have different characteristics, are evaluated by both software simulator and a prototype machine.This paper presents the results of discussing the data‐driven mechanism with advanced control in the list processing system. The profiles of parallelism for the five programs, dynamic execution rate of instructions, and the characteristics of waiting time in a matching store are described. Finally, the performance of the EM‐3 prototype is shown as well as the effectiveness of introducing the advanced contro

ISSN:0882-1666    

DOI:10.1002/scj.4690210502

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractThis paper analyzes the rough behavior of aliasing probabilities (missing faults) when the signature testing is applied to CMOS combinational circuits. In the analysis of aliasing probabilities, the fault detection probability for one‐digit test patterns—the fault detection probability—becomes important. However, for general circuits it is difficult to evaluate the fault detection probability.In this paper the circuits are restricted to CMOS combinational circuits, and it is intended for the control circuits for microprocessors. For CMOS logic LSI's, the utilization of combinational gates follows a certain specific frequency and the expected value of the fault detection probability can be calculated.First, the gate usage frequencies are examined for three types of CMOS logic LSI's. Next, the fault detection probabilities of single gates are analyzed. Based on the forementioned results, the fault detection probabilities of simplified κ‐stage combinational circuits are analyzed. As a result, the expected value of the fault detection probability for a CMOS combinational circuit is shown to be 0.4 – 0.5. Using this expected value, the approximate behaviors of the aliasing probabilities for the signature testing are presented for MISR's and single‐input LFSR's applied to CMOS combinational circuits. In particular, for single‐input LFSR's it is shown that within the range of practical use, the aliasing probabilities are

ISSN:0882-1666    

DOI:10.1002/scj.4690210503

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractCache storage control methods in multiprocessor systems have been investigated for a long time. Recently, however, the number of processors in very large computer systems continues to increase and management by conventional control methods has become inadequate. Also, the performance evaluation becomes so complicated that it is difficult to determine the best method.Therefore, this paper proposes a simple evaluation method and a cache storage control method that maintains the overhead of storage management against the increase in the processor counts. The two methods to manage the STORE instructions are store‐through and store‐in. However, there does not exist a system with two‐level cache storage of the store‐in method. This paper presents the following three goals:(1) to propose a control method that manages the situation in which the distances between storage controllers and instruction processors are not uniform;(2) to show a hardware control method that manages the shared and writable storage blocks.To achieve small overhead by the two forementioned points; and(3) to show by using a simple evaluation method that in many practical cases, the three‐level, single‐copy, store‐in method is

ISSN:0882-1666    

DOI:10.1002/scj.4690210504

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractData such as vectors, points and symbols of maps and facilities location maps are classified into layers and managed according to such attributes as roads, houses, and facilities. In editing a diagram, range searchings and neighborhood searching frequently are made, where the objects of search are the data in a number of layers. When such multidimensional data classified into layers (called multilayer data) are managed by the conventional multidimensional data structure, a problem occurs in that the retrieval efficiency varies greatly depending on the number of considered layers. This paper proposes the ML structure (multi‐layered structure), which is a new managing structure for the multilayer data based on a tree structure.In ML structure, the node structure is extended so that data can be stored not only in the terminal node of the tree but also in the internal node. A layer management mechanism is added to each node. It is arranged that the data of the layer with small number of data are placed close to the root of the tree structure. As a result, a satisfactory performance with small variation in the retrieval efficiency is realized, even if a range searching is made for a particular layer, or for a number of layers.This paper describes the construction of ML structure, method of data management, and searching. It is shown by experiment that the data management and searching by ML structure is effective for the multilayer dat

ISSN:0882-1666    

DOI:10.1002/scj.4690210505

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractBy replacing the calculation of the trigonometric functions sin and cos in the Hough transform with the piecewise‐linear Hough function (PLH), the number of multiplication and trigonometric calculations are reduced. The PLH function is derived from the standard Hough function in a straightforward manner. The PLH‐based transformation retains the Hough transform's ability to extract collinearities from an image.The cost of computing the PLH transform (PLHT) is approximately one‐sixth that of the standard Hough transform. It was also found that the PLHT reduces memory requirements by 30 pe

ISSN:0882-1666    

DOI:10.1002/scj.4690210506

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractThis paper describes a fast labeling algorithm which separates regions lying all over an image based on a pipeline method. In the previous labeling algorithm of raster‐scan type, a bottleneck for fast processing was the label classification process to classify temporarily assigned labels based on the connectivity. To cope with this problem, consideration was given to how to reduce the number of connectivities input to the labeling process, and how to make the labeling process efficient.For the problem of the number of connectivities, a method was developed for reducing the number of connectivities due to reshaping of a region, and a fast method for reducing the number of duplicate connectivities by introduction of a dividing process. Also, for the problem of label classification, a label classification method was developed using direct search for connectivities. Moreover, the relationship between the number of provisional labels in the labeling process and the number of connectivities was investigated. As a result, it was shown that the label classification process can be implemented in time proportional to the number of provisional labels and the effectiveness of the algorithm presented here was verified by simulatio

ISSN:0882-1666    

DOI:10.1002/scj.4690210507

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractThe problems with recent approaches in qualitative and causal reasoning are that: (1) in qualitative modeling, they do not discuss on what types of mathematical models their qualitative equations should be based. This allows the misuse of static models for the qualitative equations to discuss changes in the transient state; and (2) they do not define causality on their qualitative model.This paper defines the “inherent causality” which requires “time reference.” With this causality, causal reasoning is carried out by verifying that any change is made by the causality. The verification is done by consumption ofdt= + for each step. To make such causal reasoning possible, the base model (i.e., the dynamical model which describes causality from what makes the change to what is changed) is selected carefully. Also, the qualitative simulation algorithm utilizing the inherent causality is developed. The power of the causality and simulation algorithm is demonstrated on the examples of a buffer tank. AQUA, a knowledge acquisition tool for process diagnosis domain, is implemented using the qualitative simulation. AQUA uses the qualitative simulation in generating fault‐syndrome ass

ISSN:0882-1666    

DOI:10.1002/scj.4690210508

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY

摘要:

AbstractA method is presented which recovers the 3‐D shape of a polyhedron under the assumption of a large number of parallel edges. The method is based on the fact that three or more parallel edges intersect at a point (vanishing point) when extended on the image plane. To determine a pair of two parallel edges, a heuristic method based on the structure of polyhedra is employed. To cope with the error in the image, a threshold processing based on the causes of error is proposed.When parallel edges are found, the 3‐D orientation of each edge can be determined from the vanishing point. However, contradictions may arise in the calculated 3‐D orientations of the edges due to errors. To overcome such a problem, an optimization technique based on the constraint on polyhedra is proposed. Then our method is applied to an actual

ISSN:0882-1666    

DOI:10.1002/scj.4690210509

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1990

数据来源: WILEY