-
公开(公告)号:JP2000229994A
公开(公告)日:2000-08-22
申请号:JP3601899
申请日:1999-02-15
Applicant: NEC CORP
Inventor: ASOGAWA MINORU
Abstract: PROBLEM TO BE SOLVED: To provide both a method and an apparatus for estimating a three- dimensional structure of protein, capable of taking the restriction of three- dimensional structure into consideration and estimating a three-dimensional structure having a high likelihood. SOLUTION: This apparatus is equipped with a partial sequence combination generating part 11 for forming partial sequences for an amino acid sequence and generating the whole combinations of one or two partial sequences of the formed partial sequences, a bond energy calculating part 12 for calculating the bond energy of the partial sequences or the combinations of the partial sequences, a restriction calculating part 13 of three-dimensional structure for calculating the restriction of the three-dimensional structure of the partial sequences or the combinations of the partial sequences and a Hopfield neutral network part 14 for using the calculated bond energy and the calculated restriction as initial values and obtaining a combination for minimizing the total of the bond energy of the partial sequences while satisfying the restriction of three-dimensional structure.
-
公开(公告)号:JPH04373091A
公开(公告)日:1992-12-25
申请号:JP15078391
申请日:1991-06-24
Applicant: NEC CORP
Inventor: ASOGAWA MINORU
Abstract: PURPOSE:To recognize the classification of character patterns including offset, rotating or parallelly moving. CONSTITUTION:This device is composed of 5 standard character pattern/ category memory part 11 to hold a standard character pattern and the category of the character pattern, input area part 14 adaptive to the character pattern including offset, rotating or parallelly moving pattern associated neural network 12 to recognize the character pattern shown at a standard position, limiting force control part 15 to control the influence of spatial limiting force at the input area part in the case of recognition, and learning/recognition control part 13 to control the pattern associated type neural network part.
-
公开(公告)号:JPH03218731A
公开(公告)日:1991-09-26
申请号:JP1519490
申请日:1990-01-24
Applicant: NEC CORP
Inventor: ASOGAWA MINORU
IPC: A61B5/0452
Abstract: PURPOSE:To improve the recognition accuracy and the maximum likelihood of a partition point by using a pattern associative type neural network and a recurrent neural network, and executing the recognition of a partition by taking notice of an electrocardiogram waveform of that time and a timewise structure of the partition. CONSTITUTION:A leaning/recognition control part 16 fetches a value St of an electrocardiogram waveform of the present time point (t) of the electrocardiogram and its partition Kt from a standard waveform/partition pattern memory part 11, and uses them for the learning of internal parameters of a pattern associative type neural network 12 and a recurrent neural network 13. The network 12 sets a signal S't obtained by superposing the value St of the electrocardiogram waveform of the time point (t) and the output Nt of a noise generator 14, and a partition Kt-1 of the time point (t)-1 as input signals, sets the partition Kt of the time point (t) as a teacher signal, and updates the internal parameter of the network 12. The network 13 sets the partition Kt-1 of the time point (t)-1 as the input signal and sets the partition Kt of the time point (t) as the teacher signal, and updated the internal parameter of the network 13.
-
公开(公告)号:JPH02115933A
公开(公告)日:1990-04-27
申请号:JP26972788
申请日:1988-10-25
Applicant: NEC CORP
Inventor: ASOGAWA MINORU
IPC: G06F9/44
Abstract: PURPOSE:To facilitate the preparation of a rule by supporting the arrangement of knowledge of the condition part element and conclusion part element of a problem area, and the arrangement of relation between the condition part element and the conclusion part element of a rule on a rule generator. CONSTITUTION:The key word of the rule desired to generate is selected from a key word data library 24, and is registered on a key word buffer 74. A condition element is duplicated from a condition part element data library 22, and is registered on a condition part element buffer 72. Also, a conclusion element is duplicated from a conclusion part element data library 23, and is registered on a conclusion part element buffer 73. And the conclusion part element of the rule desired to generate is selected from the conclusion part element buffer 73. Also, the condition part element corresponding to the conclusion part element of the rule is selected from the condition part element buffer 72. The rule including selected condition part element and conclusion part element is registered on a rule library 92.
-
公开(公告)号:JP2004318321A
公开(公告)日:2004-11-11
申请号:JP2003109054
申请日:2003-04-14
Inventor: NAKAZATO TAKERU , MIYAGAWA TOMOYA , KENMOCHI AKIHISA , ASOGAWA MINORU
IPC: G06F17/30
CPC classification number: G06F19/28
Abstract: PROBLEM TO BE SOLVED: To provide a biological information retrieval system and method for solving a problem in retrieving a biological database that it is impossible to retrieve a gene or protein with anything other than its name.
SOLUTION: A user input accepting part 11 which has received an inquiry from a user to a biological information database generates a user query 111, and a query generating part 12 generates a biological information database query 121, and a biological information database retrieval part 13 actually retrieves a biological information database 999, and outputs a biological information database retrieval result 133. A retrieval result analysis part 14 analyzes the retrieval result, and outputs an analysis result 142, and an output control part 15 receives the analysis result, and successively outputs it, and generates a re-retrieval query 151 for non-retrieval, and transmits it to the query generating part 12. An output information designating and holding part 16 preliminarily holds conditions associated with the extraction and output of the retrieval result.
COPYRIGHT: (C)2005,JPO&NCIPI-
公开(公告)号:JP2004184137A
公开(公告)日:2004-07-02
申请号:JP2002349258
申请日:2002-11-29
Inventor: HATTORI WATARU , BABA MASAKAZU , SANO TORU , IIDA KAZUHIRO , KAWAURA HISAO , IGUCHI NORIYUKI , SOMEYA HIROKO , ASOGAWA MINORU
CPC classification number: H01J49/04 , G01N2001/4027
Abstract: PROBLEM TO BE SOLVED: To provide a chip for mass spectrometry for an LDI-TOFMS (laser desorption/ionization time-of-flight mass spectrometry) for concentrating and drying efficiently a sample, in particular, a micro sample, and a chip for mass spectrometry used as a substrate for LDI-TOFMS measurement, after drying the sample.
SOLUTION: Holes 103 are formed in the substrate 101 by etching or the like. A sample solution is dropped in the each hole 103 to provide the dried sample. The obtained dry sample is supplied to the LDI-TOFMS together with the substrate 101.
COPYRIGHT: (C)2004,JPO&NCIPI-
公开(公告)号:JP2004012355A
公开(公告)日:2004-01-15
申请号:JP2002167914
申请日:2002-06-07
Applicant: NEC CORP
Inventor: TSUDA KENICHIRO , SHIMADA JIRO , ASOGAWA MINORU , KAMIJO KENICHI , YAMAZAKI TOSHIMASA
Abstract: PROBLEM TO BE SOLVED: To provide a method and a system that can reduce the number of experiments in identification with a small amount of a sample and an experiment for trial-and-error, adds a mass spectrum that may appear, a peak or the like buried in noise, or the like for achieving a high accuracy in protein identification and high throughput, and can determine sequencing using mass spectrometry even in new protein that is not registered in a protein sequencing database. SOLUTION: In the mass spectrometry due to a tandem type mass spectrometry apparatus, bond dissociation energy is calculated by using a molecular simulation method, the prediction information of the cleavage pattern of peptide and/or the peak position of the mass spectrum and/or the intensity ratio of a spectrum is added to obtained experiment information for use as the identification of protein. COPYRIGHT: (C)2004,JPO
-
公开(公告)号:JPH04271490A
公开(公告)日:1992-09-28
申请号:JP5396591
申请日:1991-02-26
Applicant: NEC CORP
Inventor: ASOGAWA MINORU
Abstract: PURPOSE:To improve both accuracy and likelihood of character recognition for a character pattern which includes a noise, rotates or moves in parallel. CONSTITUTION:A standard character pattern category memory part 11 holds N-pairs of standard character patterns for learning and the category. A pattern associated type neural network 12 learns with a standard character pattern St as an input signal and with a category Kt as a teacher signal and recognizes the character pattern St and the category Kt. A learning and recognizing control part 13 controls the actions of the learning and recognition of the pattern associated type neural network 12. The pattern from the external part becomes the input of the pattern associated type neural network 12 through an input area part 14. A constraint force control part 15 controls the influence of the spacious constraint force of the input area part. A switching part 16 switches the input to the pattern associated type neuralnetwork 12 between phases of the learning and the recognition.
-
公开(公告)号:JPH03116389A
公开(公告)日:1991-05-17
申请号:JP25411289
申请日:1989-09-29
Applicant: NEC CORP
Inventor: ASOGAWA MINORU
IPC: A61B5/0452 , G06T7/00
Abstract: PURPOSE:To improve accuracy and exactitude for recognizing a dividing point by executing the recognition of the dividing point while paying attention to the time structure of an electrocardiogram waveform. CONSTITUTION:A standard dividing point pattern memory part 11 is provided to hold a standard part pattern and the dividing point of the waveform, and a noise generator 12 is provided to superimpose noise to the standard pattern and to increase the kind of learning patterns. Then, a recurrent neural network part 13 is provided to segment the dividing point from the learning pattern, to which the noise signal is superimposed, and a learn control part 14 is provided to control the learn of the recurrent neural network part 13. The recognition of the dividing point is executed while paying attention to the time structure of the electrocardiogram waveform. Thus, both the accuracy and the exactitude is improved for recognizing the dividing point and by executing the recurrent neural network learning based on the waveform for which the noise is superimposed to the standard pattern, only the specified learning pattern can be prevented from being over learnt.
-
公开(公告)号:JPH03114439A
公开(公告)日:1991-05-15
申请号:JP25411189
申请日:1989-09-29
Applicant: NEC CORP
Inventor: ASOGAWA MINORU
IPC: A61B5/0452 , G06T7/00
Abstract: PURPOSE:To enhance the confirmation accuracy and exactitude of a sectional point by providing a standard sectional point pattern memory part holding a standard pattern and the sectional point of the waveform thereof, a recurrent neural network part cutting out the sectional point from the standard pattern and a learning control part controlling learning. CONSTITUTION:(N) sets of a standard electrocardiogram waveform and the sectional points thereof are held to a standard sectional point pattern memory part 11 and a learning control part 13 takes out the value St of an electrocardiogram at the present point of time (t) of an electrocardiogram waveform and the sectional point Kt thereof. The sectional point Kt is set to the teacher signal of a recurrent neural network part 12 and the value St of the electrocardiogram is set to the input signal of the recurrent neural network part 12. On the basis of these input signal St and teacher signal Kt, the renewal of the internal parameter in the recurrent neural network part 12 is performed. learning operation is performed with respect to the set of the standard pattern St and sectional point Kt (1
-
-
-
-
-
-
-
-
-