Hitachi, Ltd.

Nippon Telegraph and Telephone Public Corporation

Masahiro Abe

Hiroshi Sakou

Junichi Higashino

Hisao Ogata

Katsuo Motosaka

Takehisa Hayashi

Kouki Noguchi

Tsuneya Kurihara

Masayuki Fujio

Yoshiaki Eifuku

Toshio Kajiwara

Toshiaki Sumi

A neural network (100) has an input layer, a hidden layer, and an output layer. The neural network stores weight values which operate on data input at the input layer to generate output data at the output layer. An error computing unit (87) receives the output data and compares it with desired output data from a learning data storage unit (105) to calculate error values representing the difference. An error gradient computing unit (81) calculates an error gradient, i.e. rate and direction of error change. A ratio computing unit (82) computes a ratio or percentage of a prior conjugate vector and combines the ratio with the error gradient. A conjugate vector computing unit (83) generates a present line search conjugate vector from the error gradient value and a previously calculated line search gradient vector. A line search computing unit (95) includes a weight computing unit (88) which calculates a weight correction value. The weight correction value is compared (18) with a preselected maximum or upper limit correction value (.kappa.). The line search computing unit (95) limits adjustment of the weight values stored in the neural network in accordance with the maximum weight correction value.

Learning method and apparatus for neural networks and simulator with

An automatic pattern recognition method has a short processing time, can be applied to nonlinear separation problems, and can perform similarity calculations. The method: Divides a plurality of sample data of known categories into a plurality of classes; When the sample data in a divided class is not all in the same category, repeats dividing the sample data into subclasses until sample data in a subclass has only one category; Expresses the relationship between classes and subclasses in a tree-structure representation and determines the standard pattern for each class and subclass from the sample data contained there; and Checks which of the tree-structured classes input data of unknown category is nearest, by calculating the distance to the standard pattern of each class, and then, when the class has subclasses, performs a similar check until the lowest-level subclass is reached to determine the subclass the input data is closest to. The category of the lowest-level subclass is taken as the category of the input data.

Automatic clustering method

A learning control method reduces overall learning time by displaying data related to an appropriate determination of learning protraction and a proper restoring method. Prior to initiating the learning, the user is inquired about the current problem and a problem data set representing items associated with the problem is obtained. Evaluation data indicating a state of learning obtained during the learning on the current problem is sequentially stored and displayed. When there is a high possibility of learning protraction during the learning, a message informing the user is displayed. When the learning is stopped by the user in this case, the problem data set and evaluation data set are stored. Then, a list of restoring methods is displayed and a particular restoring method is selected by the user once the learning is stopped. The learning is restarted on the current problem in accordance with the selected restoring method.

Method of and system for controlling learning in neural network

A document processing system for processing documents by using structured keywords comprises an output system and a receiver system. The output system includes a first storage for storing a structured keyword dictionary containing structured keywords among which relations are systematically structured, and linkage unit providing linkage information for establishing correspondences between constituent parts of an input document and corresponding ones of the keywords. The receiver system is coupled to the output system and includes a second storage for storing structured keywords among which relations are systematically structured, and retrieving unit having inputs supplied with the document and the linkage information for retrieving the document to thereby form data of a predetermined edition format by using the structured keyword read out from the second storage. Data transfer between the output system and the receiver systems can be performed either on-line or off-line.

Information processing system and method for processing document by

Method and system of predicting attribute value in the learning system, (a) where expected value of equivalent example and its classification class is generated by domain knowledge, with respect to the test example being supplemented with candidates of a plural number of attribute values, in which the attribute lacking part of attribute values can adopt in the learning system, having (i) attribute descriptions containing a plural number of training examples, each of which contains attribute and its attribute value, and such training example and classification class adopted by each training example, and (ii) domain knowledge generating a pair of second example and its classification class from a pair of first example and its classification class, as equivalent example, (b) where classification class of equivalent example generated as above is predicted by decision tree produced based on the above attribute descriptions, and (c) where the attribute value lacking the above test example is predicted by use of the expected value of classification class and the above prediction result with respect to the above equivalent example.

Attribute value predicting method and system in the learning system

A vortex blower having an impeller provided with an annular recess laterally opening around an axis of rotation, a casing disposed so as to face the impeller and cover a radially outer portion of the impeller, an arcuate recess formed around the rotation axis in a portion of the casing which extends between a suction port and a discharge port and opposes to the annular recess, and a plurality of blades provided in the annular recess to divide the recess into a plurality of circumferentially arranged pockets, each having a radially outer end extending towards the arcuate recess beyond the annular recess. The vortex blower having such a construction can operate safely even when drawing air in which dust is suspended.

Vortex blower

A vortex blower including a casing, an annular air passageway defined in the casing, an impeller mounted for rotation in the casing, a partition wall, a suction port and a discharge port disposed on both sides of the partition wall and communicated with each other through the annular air passageway, and an extension wall formed by extending at least one wall surface of the partition wall along a surface of rotation of the impeller, the extension wall tapering in going toward its forward end. The blower is further formed with at least one projecting wall continuous with the partition wall and an inner wall surface of the air passageway in the vicinity of the partition wall, for shutting off the noise produced in the vicinity of the partition wall and preventing it from being transmitted to outside to thereby reduce the noise produced by the blower.

A vortex blower comprises a motor and an impeller directly connected to one end of a motor shaft and housed in an impeller housing. A part of the impeller housing is made of aluminium or aluminium-based alloy and includes an integral portion which serves as that end wall of the motor which is adjacent to the impeller. The end wall is provided with a radially extending first thrust surface for engagement with one end face of an outer race of a bearing which journals the motor shaft. A bearing retainer of a metal having a wear-resistant property higher than those of aluminium and the aluminium-based alloy is provided with an axially extending cylindrical surface for engagement with the outer peripheral surface of the outer bearing race and a radially extending second thrust surface for engagement with the other end face of the outer bearing race. The bearing retainer is mounted on the end wall so that the outer peripheral surface and end faces of the outer bearing race are engaged with the cylindrical surface of the bearing retainer and the first and second thrust surfaces, respectively. The use of the bearing retainer fabricated separately of the end wall and mounted thereon eliminates the use of an insert which was required in the prior art and thus simplifies the steps of manufacture and reduces the number of necessary machine parts.

A communication control system comprises a communication control unit and a line control unit coupled to it. The former communicates with a data processor and the latter communicates with a plurality of modems. A first address signal generator in the communication control unit sends an address signal to a first decoder in the line control unit, allowing it to select one of the line adapters in the line control unit and, therefore, one of the modems coupled to the line adapter. The selected line adapter sends a signal and data if any to the communication control unit, which send a signal and data, if any, to the line control unit in response to a signal or data sent from the line control unit. A second address signal generator in the communication control unit begins to send a little later the second address signal which is the same as the first address signal to a second decoder in the line control unit, allowing it to select the same one of the line adapters as one previously selected by the first decoder. The line adapter selected by the second decoder can receive a signal and data if any, sent from the communication control unit. At the time when the signal and data are received by the line adapter selected by the first decoder, the first decoder has selected the next line adapter, because the next first address has already been sent by the first address generator. Thus, the communication control unit selects two of the line adapters at the same time allowing one to send signal and data to the communication control unit and allowing the other to receive signal and data sent from the communication control unit in response to the signal and data already sent from the line control unit.

Growing community of inventors

Iruma, Japan

Masahiro Abe