中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

10 records (took 0.018 seconds)

    Model file management method and terminal device
    1.
    发明授权
    Model file management method and terminal device 有权

    公开(公告)号：US11940992B2

    公开(公告)日：2024-03-26

    申请号：US17284898

    申请日：2019-11-01

    Applicant: Huawei Technologies Co., Ltd.

    Inventor： Qing Su ,  Junyuan Yang ,  Wenshuai Yin ,  Yue Gu ,  Ke Wan

    IPC: G06F16/23 ,  H04L67/06 ,  G06F8/61 ,  H04L67/00

    CPC classification number: G06F16/2379 ,  H04L67/06 ,  G06F8/61 ,  H04L67/34

    Abstract: A model file management method includes that a terminal device receives a storage address of a target model file package from a server and the terminal device obtains the target model file package based on the storage address of the target model file package, where the target model file package is based on a parameter of a model file package locally stored in the terminal device and a parameter of a model file package managed by the server. In an artificial intelligence (AI) field, an application may implement a specific function by using an AI model file. An application is decoupled from an AI model file such that the terminal device performs centralized management on a general model file.

    Method, apparatus, and system for operating shared resource in asynchronous multiprocessing system
    2.
    发明授权
    Method, apparatus, and system for operating shared resource in asynchronous multiprocessing system 有权
    在异步多处理系统中运行共享资源的方法，装置和系统

    公开(公告)号：US09519652B2

    公开(公告)日：2016-12-13

    申请号：US13870586

    申请日：2013-04-25

    Applicant: Huawei Technologies Co., Ltd.

    Inventor： Qing Su ,  Yu Liu ,  Xiancheng Guo ,  Hu Liu

    IPC: G06F17/30 ,  G06F13/16

    CPC classification number: G06F17/30171 ,  G06F13/1663

    Abstract: Technical effects of a method, an apparatus, and a system for operating a shared resource in an asynchronous multiprocessing system that are provided in the present invention are as follows: A processor in an asynchronous multiprocessing system implements an operation on a shared resource by locking a hardware resource lock, and the hardware resource lock is implemented by a register; in this way, a bus in the asynchronous multiprocessing system does not need to support a synchronization operation, and the processor also does not need to have a feature of supporting a synchronization operation, and is capable of implementing the operation on the shared resource only in a manner of accessing the register, which simplifies the operation on the shared resource by the processor in the asynchronous multiprocessing system, enlarges a selection range of the processor in the asynchronous multiprocessing system, and further improves flexibility of the asynchronous multiprocessing system.

    Abstract translation: 在本发明中提供的用于在异步多处理系统中操作共享资源的方法，装置和系统的技术效果如下：异步多处理系统中的处理器通过锁定共享资源来实现对共享资源的操作 硬件资源锁定，硬件资源锁定由寄存器实现; 以这种方式，异步多处理系统中的总线不需要支持同步操作，并且处理器也不需要具有支持同步操作的特征，并且能够仅在共享资源中实现对共享资源的操作 访问寄存器的方式简化了异步多处理系统中的处理器对共享资源的操作，扩大了异步多处理系统中的处理器的选择范围，并进一步提高了异步多处理系统的灵活性。

    Method for training neural network model and apparatus
    4.
    发明授权
    Method for training neural network model and apparatus 有权

    公开(公告)号：US11521012B2

    公开(公告)日：2022-12-06

    申请号：US16910289

    申请日：2020-06-24

    Applicant: HUAWEI TECHNOLOGIES CO., LTD.

    Inventor： Tao Ma ,  Qing Su ,  Ying Jin

    IPC: G06N3/08 ,  G06K9/62 ,  G06F8/71 ,  G06N3/04 ,  G06N5/04

    Abstract: This application provides a method for training a neural network model and an apparatus. The method includes: obtaining annotation data that is of a service and that is generated by a terminal device in a specified period; training a second neural network model by using the annotation data that is of the service and that is generated in the specified period, to obtain a trained second neural network model; and updating a first neural network model based on the trained second neural network model. In the method, training is performed based on the annotation data generated by the terminal device, so that in an updated first neural network model compared with a universal model, an inference result has a higher confidence level, and a personalized requirement of a user can be better met.

    METHOD, APPARATUS, AND SYSTEM FOR OPERATING SHARED RESOURCE IN ASYNCHRONOUS MULTIPROCESSING SYSTEM
    5.
    发明申请
    METHOD, APPARATUS, AND SYSTEM FOR OPERATING SHARED RESOURCE IN ASYNCHRONOUS MULTIPROCESSING SYSTEM 有权
    在异步多媒体系统中操作共享资源的方法，装置和系统

    公开(公告)号：US20130290286A1

    公开(公告)日：2013-10-31

    申请号：US13870586

    申请日：2013-04-25

    Applicant: Huawei Technologies Co., Ltd.

    Inventor： Qing Su ,  Yu Liu ,  Xiancheng Guo ,  Hu Liu

    IPC: G06F17/30

    CPC classification number: G06F17/30171 ,  G06F13/1663

    Abstract: Technical effects of a method, an apparatus, and a system for operating a shared resource in an asynchronous multiprocessing system that are provided in the present invention are as follows: A processor in an asynchronous multiprocessing system implements an operation on a shared resource by locking a hardware resource lock, and the hardware resource lock is implemented by a register; in this way, a bus in the asynchronous multiprocessing system does not need to support a synchronization operation, and the processor also does not need to have a feature of supporting a synchronization operation, and is capable of implementing the operation on the shared resource only in a manner of accessing the register, which simplifies the operation on the shared resource by the processor in the asynchronous multiprocessing system, enlarges a selection range of the processor in the asynchronous multiprocessing system, and further improves flexibility of the asynchronous multiprocessing system.

    Abstract translation: 在本发明中提供的用于在异步多处理系统中操作共享资源的方法，装置和系统的技术效果如下：异步多处理系统中的处理器通过锁定共享资源来实现对共享资源的操作 硬件资源锁定，硬件资源锁定由寄存器实现; 以这种方式，异步多处理系统中的总线不需要支持同步操作，并且处理器也不需要具有支持同步操作的特征，并且能够仅在共享资源中实现对共享资源的操作 访问寄存器的方式简化了异步多处理系统中的处理器对共享资源的操作，扩大了异步多处理系统中的处理器的选择范围，并进一步提高了异步多处理系统的灵活性。

    Model File Management Method and Terminal Device
    6.
    发明申请
    Model File Management Method and Terminal Device 有权

    公开(公告)号：US20210390092A1

    公开(公告)日：2021-12-16

    申请号：US17284898

    申请日：2019-11-01

    Applicant: Huawei Technologies Co., Ltd.

    Inventor： Qing Su ,  Junyuan Yang ,  Wenshuai Yin ,  Yue Gu ,  Ke Wan

    IPC: G06F16/23 ,  H04L29/08

    Abstract: A model file management method includes that a terminal device receives a storage address of a target model file package from a server and the terminal device obtains the target model file package based on the storage address of the target model file package, where the target model file package is based on a parameter of a model file package locally stored in the terminal device and a parameter of a model file package managed by the server. In an artificial intelligence (AI) field, an application may implement a specific function by using an AI model file. An application is decoupled from an AI model file such that the terminal device performs centralized management on a general model file.

    Tree Topology Based Computing System and Method
    7.
    发明申请
    Tree Topology Based Computing System and Method 审中-公开

    公开(公告)号：US20200342297A1

    公开(公告)日：2020-10-29

    申请号：US16926121

    申请日：2020-07-10

    Applicant: Huawei Technologies Co., Ltd.

    Inventor： Mingyang Dai ,  Qing Su ,  Xiaofei Wang

    IPC: G06N3/063 ,  G06K9/62 ,  G06F16/2455

    Abstract: A tree topology based computing system and method, where the system may include a plurality of node clusters, where the plurality of node clusters constitute a multi-layer network structure in a tree topology manner, any minimum tree in the network structure includes a second node cluster and at least one first node cluster. The first node cluster is configured to obtain a first computing result based on a first computing input, and send the first computing result to the second node cluster, and the second node cluster is configured to receive at least one first computing result sent by the at least one first node cluster, and aggregate the at least one first computing result and a second computing result to obtain a third computing result.

    TEXT READING METHOD AND DEVICE
    9.
    发明公开
    TEXT READING METHOD AND DEVICE 审中-公开

    公开(公告)号：US20240194182A1

    公开(公告)日：2024-06-13

    申请号：US18556656

    申请日：2022-03-31

    Applicant: HUAWEI TECHNOLOGIES CO., LTD.

    Inventor： Qing Su ,  Feng Sun ,  Chu Yang ,  Lin Cao ,  Xiaoyue Huo ,  Hongfeng Ni

    IPC: G10L13/08 ,  G06F3/0488 ,  G10L13/047

    CPC classification number: G10L13/08 ,  G06F3/0488 ,  G10L13/047

    Abstract: An electronic device displays a first user interface; receives a first operation of a user; obtains first content of the first user interface in response to the first operation; displays a second user interface, where content displayed in the second user interface includes a text in the first content, and the second user interface covers a part of a display area of the first user interface; reads a first sentence in the content of the second user interface; and displays marking information of a text that is in the second user interface and that corresponds to the first sentence that is being read. Embodiments of this application are used for text reading.

    METHOD FOR TRAINING NEURAL NETWORK MODEL AND APPARATUS
    10.
    发明申请
    METHOD FOR TRAINING NEURAL NETWORK MODEL AND APPARATUS 审中-公开

    公开(公告)号：US20200320344A1

    公开(公告)日：2020-10-08

    申请号：US16910289

    申请日：2020-06-24

    Applicant: HUAWEI TECHNOLOGIES CO., LTD.

    Inventor： Tao Ma ,  Qing Su ,  Ying Jin

    IPC: G06K9/62 ,  G06N3/04 ,  G06F8/71 ,  G06N5/04 ,  G06N3/08

    Abstract: This application provides a method for training a neural network model and an apparatus. The method includes: obtaining annotation data that is of a service and that is generated by a terminal device in a specified period; training a second neural network model by using the annotation data that is of the service and that is generated in the specified period, to obtain a trained second neural network model; and updating a first neural network model based on the trained second neural network model. In the method, training is performed based on the annotation data generated by the terminal device, so that in an updated first neural network model compared with a universal model, an inference result has a higher confidence level, and a personalized requirement of a user can be better met.

Patent Agency Ranking