知搜-全球专利检索

  1. Login
  2. Sign Up

Search Results

17 records (took 0.018 seconds)

    LEARNING DEVICE, LEARNING METHOD, AND COMPUTER-READABLE MEDIUM
    1.
    发明公开
    LEARNING DEVICE, LEARNING METHOD, AND COMPUTER-READABLE MEDIUM 审中-公开

    公开(公告)号：US20230143070A1

    公开(公告)日：2023-05-11

    申请号：US17619723

    申请日：2021-03-10

    Applicant: NEC Corporation

    Inventor： Takaya MIYAMOTO ,  Hiroshi HASHIMOTO

    IPC: G06N20/00

    CPC classification number: G06N20/00

    Abstract: A learning device (12) includes: an input unit (109) that inputs target data to be learned, class label information of the target data, and statistical property information of the target data; a feature amount extractor (110) that extracts a feature amount from the target data by using a parameter; a class classifier (111) that outputs a class classification inference result of the target data by statistical processing using the feature amount and a weight vector of each class; a loss calculation unit (112) that calculates a loss by using a loss function in which the class classification inference result and the class label information are taken as inputs; and a parameter correction unit (113) that corrects the weight vector of the class classifier (111) and the parameter of the feature amount extractor (110) in such a way as to reduce the loss, according to the statistical property information.

    NEURAL NETWORK RANK OPTIMIZATION DEVICE AND OPTIMIZATION METHOD
    2.
    发明申请
    NEURAL NETWORK RANK OPTIMIZATION DEVICE AND OPTIMIZATION METHOD 有权

    公开(公告)号：US20210073633A1

    公开(公告)日：2021-03-11

    申请号：US16963377

    申请日：2018-10-24

    Applicant: NEC Corporation

    Inventor： Hiroshi HASHIMOTO

    IPC: G06N3/08 ,  G06N3/04 ,  G06N3/063 ,  G06F17/16

    Abstract: A neural network rank optimization device includes tensor decomposition means 1 for performing tensor decomposition processing of tensor-decomposing a convolution filter included in a neural network for low rank approximation, and rank optimization means 2A for optimizing a rank used in the tensor decomposition means 1, wherein the rank optimization means includes evaluation amount calculation means 3 for calculating an evaluation amount representing the degree of the low rank approximation, and rank determination means 4 for determining a rank corresponding to the evaluation amount less than a predetermined threshold value as a desired rank.

    CUSTOMER ANALYSIS APPARATUS, CUSTOMER ANALYSIS METHOD, AND NON-TRANSITORY STORAGE MEDIUM
    4.
    发明申请
    CUSTOMER ANALYSIS APPARATUS, CUSTOMER ANALYSIS METHOD, AND NON-TRANSITORY STORAGE MEDIUM 有权

    公开(公告)号：US20230080815A1

    公开(公告)日：2023-03-16

    申请号：US17801936

    申请日：2020-02-28

    Applicant: NEC Corporation

    Inventor： Makiko AKIGUCHI ,  Azusa FURUKAWA ,  Kei SHIBUYA ,  Kan ARAI ,  Hiroshi HASHIMOTO ,  Ken HANAZAWA

    IPC: G06V40/16 ,  G06T11/00 ,  G06V20/52 ,  G06T7/246 ,  G06T7/292 ,  G06Q30/02

    Abstract: A customer analysis apparatus (10) includes an image acquisition unit (110) and an image processing unit (120). The image acquisition unit (110) acquires a first image generated by a first image capture unit (22) and a second image generated by a second image capture unit (24). The image processing unit (120) processes the first image and the second image. More specifically, the image processing unit (120) generates product information indicating a product (50) taken out from a product shelf (40), by processing the first image and the second image. Further, the image processing unit (120) generates customer attribute information of a customer taking out the product (50), by processing the first image. Then, the image processing unit (120) causes a result storage unit (130) to store the product information and the customer attribute information in association with each other.

    INFORMATION PROCESSING APPARATUS, INFORMATION PROCESSING METHOD, AND NON-TRANSITORY RECORDING MEDIUM
    7.
    发明申请
    INFORMATION PROCESSING APPARATUS, INFORMATION PROCESSING METHOD, AND NON-TRANSITORY RECORDING MEDIUM 有权

    公开(公告)号：US20250117464A1

    公开(公告)日：2025-04-10

    申请号：US18729982

    申请日：2022-03-16

    Applicant: NEC Corporation

    Inventor： Hiroshi HASHIMOTO ,  Akihiro HAYASAKA

    IPC: G06F21/31

    Abstract: An information processing apparatus includes: a first score acquisition unit that performs one-to-N verification by using target data acquired from a verification target and N pieces of registration data registered in advance, thereby acquiring a first matching score; and a second score acquisition unit that transforms the first matching score on the basis of an amount depending on a number of verifications N of the registration data, thereby acquiring a second matching score. According to such an information processing apparatus, it is possible to transform a score of the one-to-N verification into an appropriate value.

    IMAGE GENERATION SYSTEM, IMAGE GENERATION METHOD, AND RECORDING MEDIUM
    10.
    发明公开
    IMAGE GENERATION SYSTEM, IMAGE GENERATION METHOD, AND RECORDING MEDIUM 审中-公开

    公开(公告)号：US20240020873A1

    公开(公告)日：2024-01-18

    申请号：US18026479

    申请日：2020-09-25

    Applicant: NEC Corporation

    Inventor： Hiroshi HASHIMOTO

    IPC: G06T7/73 ,  G06T5/00 ,  G06V10/776

    CPC classification number: G06T7/73 ,  G06T5/001 ,  G06V10/776 ,  G06T2207/30201 ,  G06V40/165

    Abstract: An image generation system includes: a detection unit that detects a position information about a position of a face or a position of a feature point of the face from an image; an acquisition unit that obtains a perturbed position information that takes into account an error that occurs when the position information is detected, for the position information; and a generation unit that generates a new image including the face on the basis of the perturbed position information. According to such an image generation system, it is possible to properly generate a new image in view of the error in position estimation.

Patent Agency Ranking