Patents by Inventor Yusuke Hida

Yusuke Hida has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Patent number: 12111851
    Abstract: A method may include obtaining position data and associated time entries for multiple individuals. The method may also include, for each of the individuals, associating instances of the position data with categories of locations to generate a context vector, where the context vector includes text strings describing a context of a given individual. The method may also include, for each of the individuals, squeezing the context vector to combine consecutive categories of locations to generate a count vector, where the count vector includes the text strings and a count of the consecutive categories of locations. The method may also include classifying each of the individuals based on a comparison of the count vector to a known class of individuals, and adjusting at least one factor directing flow of multiple people in a different manner based on the classification.
    Type: Grant
    Filed: March 22, 2023
    Date of Patent: October 8, 2024
    Assignee: Fujitsu Limited
    Inventor: Yusuke Hida
  • Publication number: 20240320241
    Abstract: A method may include obtaining position data and associated time entries for multiple individuals. The method may also include, for each of the individuals, associating instances of the position data with categories of locations to generate a context vector, where the context vector includes text strings describing a context of a given individual. The method may also include, for each of the individuals, squeezing the context vector to combine consecutive categories of locations to generate a count vector, where the count vector includes the text strings and a count of the consecutive categories of locations. The method may also include classifying each of the individuals based on a comparison of the count vector to a known class of individuals, and adjusting at least one factor directing flow of multiple people in a different manner based on the classification.
    Type: Application
    Filed: March 22, 2023
    Publication date: September 26, 2024
    Applicant: Fujitsu Limited
    Inventor: Yusuke HIDA
  • Patent number: 12100199
    Abstract: A computing apparatus to classify anomalies in images, by unsupervised anomaly detection on an input dataset of the images to detect anomaly portions from said images to generate, for an image in the dataset, a corresponding mask image transmitting a detected anomaly portion in the image and blocking anomaly-free portions; train a classifier ANN, including, in a first epoch process processing a masked version of the input dataset with the classifier ANN, the masked version including the image of the input dataset masked by the corresponding mask image, and training the classifier ANN to classify anomaly portions into one of plural classes by minimising a cross entropy loss function using generated labels as ground truths; extracting, from the classifier ANN, a latent feature representation of the image of the masked version in the input dataset; and in a second epoch process generating a set of pseudo labels corresponding to the masked version of the input dataset by applying an unsupervised clustering algori
    Type: Grant
    Filed: December 10, 2021
    Date of Patent: September 24, 2024
    Assignee: FUJITSU LIMITED
    Inventor: Yusuke Hida
  • Patent number: 12072323
    Abstract: An analyzer configured to acquire a chromatogram or spectrum by performing a predetermined analysis of a sample and perform a qualitative or quantitative analysis of components contained in the sample. The analyzer includes: a peak detection unit configured, based on information regarding a plurality of target components that need to be checked whether contained in the sample or that need to be quantified, to detect a peak or peaks in the chromatogram or spectrum acquired by the predetermined analysis of the sample corresponding to one of the target components, configured to acquire peak information regarding each of the peak or peaks, and configured to obtain confidence information for each of the peak or peaks, the confidence information being an indicative value of certainty of detecting a peak; and a display processing unit configured to display on a display unit a list of at least a part of the target components.
    Type: Grant
    Filed: May 8, 2019
    Date of Patent: August 27, 2024
    Assignees: SHIMADZU CORPORATION, FUJITSU LIMITED
    Inventors: Yohei Yamada, Shinji Kanazawa, Hiroyuki Yasuda, Akihiro Kunisawa, Yuzi Kanazawa, Yusuke Hida
  • Patent number: 11727670
    Abstract: A computer implemented method including acquiring a live image of a subject physical sample of a product or material; inputting the live image to a trained generator neural network to generate a defect-free reconstruction of the live image; comparing the defect-free reconstruction of the live image with the live image to determine a difference; and identifying a defect corresponding to the subject physical sample at a location of the determined difference.
    Type: Grant
    Filed: April 16, 2021
    Date of Patent: August 15, 2023
    Assignee: FUJITSU LIMITED
    Inventor: Yusuke Hida
  • Patent number: 11486866
    Abstract: When chromatogram data for a target sample have been acquired, a peak position estimator determines an estimated result of the position of the starting and/or ending point of a peak as well as the confidence value representing the reliability of the estimation, using a trained model stored in the trained model storage section. Normally, a plurality of estimated results of the starting point and/or ending point of the peak are acquired for one peak. A peak information correction processor identifies a candidate having the highest confidence as a prime candidate, and superposes a plurality of candidates including the prime candidate, with their respective confidence values, on a displayed chromatogram. An operator referring to the confidence values selects a peak which needs close checking or correction, and corrects the starting point and/or ending point of the selected peak, for example, by selecting and indicating a candidate other than the prime candidate.
    Type: Grant
    Filed: November 9, 2017
    Date of Patent: November 1, 2022
    Assignee: SHIMADZU CORPORATION
    Inventors: Takeshi Osoekawa, Yusuke Hida, Yuzi Kanazawa, Shinji Kanazawa, Yohei Yamada, Hiroyuki Yasuda, Akihiro Kunisawa
  • Publication number: 20220262108
    Abstract: A computing apparatus to classify anomalies in images, by unsupervised anomaly detection on an input dataset of the images to detect anomaly portions from said images to generate, for an image in the dataset, a corresponding mask image transmitting a detected anomaly portion in the image and blocking anomaly-free portions; train a classifier ANN, including, in a first epoch process processing a masked version of the input dataset with the classifier ANN, the masked version including the image of the input dataset masked by the corresponding mask image, and training the classifier ANN to classify anomaly portions into one of plural classes by minimising a cross entropy loss function using generated labels as ground truths; extracting, from the classifier ANN, a latent feature representation of the image of the masked version in the input dataset; and in a second epoch process generating a set of pseudo labels corresponding to the masked version of the input dataset by applying an unsupervised clustering algori
    Type: Application
    Filed: December 10, 2021
    Publication date: August 18, 2022
    Applicant: FUJITSU LIMITED
    Inventor: Yusuke HIDA
  • Publication number: 20220196615
    Abstract: An analyzer configured to acquire a chromatogram or spectrum by performing a predetermined analysis of a sample and perform a qualitative or quantitative analysis of components contained in the sample. The analyzer includes: a peak detection unit configured, based on information regarding a plurality of target components that need to be checked whether contained in the sample or that need to be quantified, to detect a peak or peaks in the chromatogram or spectrum acquired by the predetermined analysis of the sample corresponding to one of the target components, configured to acquire peak information regarding each of the peak or peaks, and configured to obtain confidence information for each of the peak or peaks, the confidence information being an indicative value of certainty of detecting a peak; and a display processing unit configured to display on a display unit a list of at least a part of the target components.
    Type: Application
    Filed: May 8, 2019
    Publication date: June 23, 2022
    Applicants: SHIMADZU CORPORATION, FUJITSU LIMITED
    Inventors: Yohei YAMADA, Shinji KANAZAWA, Hiroyuki YASUDA, Akihiro KUNISAWA, Yuzi KANAZAWA, Yusuke HIDA
  • Patent number: 11302039
    Abstract: A model constructed by a training process using the technique of deep learning using the training data including images created from a large number of chromatograms and correct peak information is previously stored in a trained model storage section. When chromatogram data for a target sample acquired with an LC measurement unit are inputted, an image creator converts the chromatogram into an image and creates an input image in which one of the two areas divided by the chromatogram curve as the boundary in the image is filled. A peak position estimator inputs the pixel values of the input image into a trained model using a neural network, and obtains the position information of the starting point and/or ending point of the peak and a peak detection confidence as the output. A peak determiner determines the starting point and/or ending point of each peak based on the peak detection confidence.
    Type: Grant
    Filed: November 9, 2017
    Date of Patent: April 12, 2022
    Assignee: SHIMADZU CORPORATION
    Inventors: Takeshi Osoekawa, Yusuke Hida, Yuzi Kanazawa, Shinji Kanazawa, Yohei Yamada, Hiroyuki Yasuda, Akihiro Kunisawa, Hidetoshi Terada
  • Patent number: 11263258
    Abstract: An information processing method implemented by a computer, the information processing method includes: acquiring an image group generated by imaging a data group according to each of a plurality of imaging methods; for each of the acquired image groups, calculating a score of the imaging method used to generate the image group, based on distribution of a first feature value group in a feature value space, and distribution of a second feature value group in the feature value space, the first feature value group being a plurality of feature values output when the image group is input to a trained model outputting feature values corresponding to input images, the second feature value group being a plurality of feature values output when a reference image group is input to the trained model; and outputting the score of the imaging method, the score being calculated for each of the image groups.
    Type: Grant
    Filed: March 6, 2020
    Date of Patent: March 1, 2022
    Assignee: FUJITSU LIMITED
    Inventors: Takashi Miura, Yusuke Hida, Atsuo Hara, Kyoko Tadaki
  • Publication number: 20210374928
    Abstract: A computer implemented method including acquiring a live image of a subject physical sample of a product or material; inputting the live image to a trained generator neural network to generate a defect-free reconstruction of the live image; comparing the defect-free reconstruction of the live image with the live image to determine a difference; and identifying a defect corresponding to the subject physical sample at a location of the determined difference.
    Type: Application
    Filed: April 16, 2021
    Publication date: December 2, 2021
    Applicant: FUJITSU LIMITED
    Inventor: Yusuke HIDA
  • Patent number: 11176455
    Abstract: A learning data generation apparatus includes a memory and a processor configured to perform determination of a region of interest in each of a plurality of images related to a learning target for machine learning in accordance with a result of image matching between the plurality of images, apply an obscuring processing to a specific region other than the region of interest in each of the plurality of images, and generate learning data including the plurality of images to which the obscuring processing is applied.
    Type: Grant
    Filed: November 25, 2019
    Date of Patent: November 16, 2021
    Assignee: FUJITSU LIMITED
    Inventor: Yusuke Hida
  • Publication number: 20200293570
    Abstract: An information processing method implemented by a computer, the information processing method includes: acquiring an image group generated by imaging a data group according to each of a plurality of imaging methods; for each of the acquired image groups, calculating a score of the imaging method used to generate the image group, based on distribution of a first feature value group in a feature value space, and distribution of a second feature value group in the feature value space, the first feature value group being a plurality of feature values output when the image group is input to a trained model outputting feature values corresponding to input images, the second feature value group being a plurality of feature values output when a reference image group is input to the trained model; and outputting the score of the imaging method, the score being calculated for each of the image groups.
    Type: Application
    Filed: March 6, 2020
    Publication date: September 17, 2020
    Applicant: FUJITSU LIMITED
    Inventors: TAKASHI MIURA, Yusuke Hida, Atsuo HARA, Kyoko Tadaki
  • Publication number: 20200292509
    Abstract: When chromatogram data for a target sample have been acquired, a peak position estimator determines an estimated result of the position of the starting and/or ending point of a peak as well as the confidence value representing the reliability of the estimation, using a trained model stored in the trained model storage section. Normally, a plurality of estimated results of the starting point and/or ending point of the peak are acquired for one peak. A peak information correction processor identifies a candidate having the highest confidence as a prime candidate, and superposes a plurality of candidates including the prime candidate, with their respective confidence values, on a displayed chromatogram. An operator referring to the confidence values selects a peak which needs close checking or correction, and corrects the starting point and/or ending point of the selected peak, for example, by selecting and indicating a candidate other than the prime candidate.
    Type: Application
    Filed: November 9, 2017
    Publication date: September 17, 2020
    Applicant: SHIMADZU CORPORATION
    Inventors: Takeshi OSOEKAWA, Yusuke HIDA, Yuzi KANAZAWA, Shinji KANAZAWA, Yohei YAMADA, Hiroyuki YASUDA, Akihiro KUNISAWA
  • Publication number: 20200284640
    Abstract: A non-transitory computer-readable recording medium records a measurement control program that controls a sensor that measures a water level, the measurement control program causing a computer to execute processing comprising: specifying a predicted peak value of the water level based on a prediction result for a change in the water level; calculating a reference value of the water level at which measurement of the water level is started, according to the predicted peak value that has been specified; specifying a timing at which the water level is predicted to reach the reference value that has been calculated, based on the prediction result; and controlling a start timing of measurement of the water level by the sensor based on the timing that has been specified.
    Type: Application
    Filed: May 26, 2020
    Publication date: September 10, 2020
    Applicant: FUJITSU LIMITED
    Inventors: Hiroshi Chiba, Takashi Suzuki, Yusuke Hida, YUTAKA NAKAGAWA
  • Publication number: 20200279408
    Abstract: A model constructed by a training process using the technique of deep learning using the training data including images created from a large number of chromatograms and correct peak information is previously stored in a trained model storage section. When chromatogram data for a target sample acquired with an LC measurement unit are inputted, an image creator converts the chromatogram into an image and creates an input image in which one of the two areas divided by the chromatogram curve as the boundary in the image is filled. A peak position estimator inputs the pixel values of the input image into a trained model using a neural network, and obtains the position information of the starting point and/or ending point of the peak and a peak detection confidence as the output. A peak determiner determines the starting point and/or ending point of each peak based on the peak detection confidence.
    Type: Application
    Filed: November 9, 2017
    Publication date: September 3, 2020
    Applicant: SHIMADZU CORPORATION
    Inventors: Takeshi OSOEKAWA, Yusuke HIDA, Yuzi KANAZAWA, Shinji KANAZAWA, Yohei YAMADA, Hiroyuki YASUDA, Akihiro KUNISAWA, Hidetoshi TERADA
  • Publication number: 20200193329
    Abstract: A computer-implemented learning method includes inputting a plurality of pieces of input data and labels representing the plurality of pieces of input data into an encoder configured to output context variables associated with each of the plurality of pieces of input data, inputting the plurality of pieces of input data and the context variables output by the encoder into a decoder configured to output decision labels associated with the plurality of pieces of input data respectively, and learning parameters of the encoder and the decoder so that each of the decision labels matches with a corresponding label of the labels representing the plurality of the plurality of pieces of input data.
    Type: Application
    Filed: December 13, 2019
    Publication date: June 18, 2020
    Applicant: FUJITSU LIMITED
    Inventors: Takeshi OSOEKAWA, TAKASHI KATOH, Yusuke Hida, Yuzi KANAZAWA
  • Publication number: 20200175366
    Abstract: A learning data generation apparatus includes a memory and a processor configured to perform determination of a region of interest in each of a plurality of images related to a learning target for machine learning in accordance with a result of image matching between the plurality of images, apply an obscuring processing to a specific region other than the region of interest in each of the plurality of images, and generate learning data including the plurality of images to which the obscuring processing is applied.
    Type: Application
    Filed: November 25, 2019
    Publication date: June 4, 2020
    Applicant: FUJITSU LIMITED
    Inventor: Yusuke Hida