中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

304 records (took 0.023 seconds)

    GRADIENT GROUPING FOR COMPRESSION IN FEDERATED LEARNING FOR MACHINE LEARNING MODELS
    21.
    发明公开
    GRADIENT GROUPING FOR COMPRESSION IN FEDERATED LEARNING FOR MACHINE LEARNING MODELS 审中-公开

    公开(公告)号：US20230325652A1

    公开(公告)日：2023-10-12

    申请号：US17714884

    申请日：2022-04-06

    Applicant: QUALCOMM Incorporated

    Inventor： Eren BALEVI ,  Taesang YOO

    IPC: G06N3/08 ,  H04W74/08

    CPC classification number: G06N3/08 ,  H04W74/0833

    Abstract: A method of wireless communication, by a user equipment (UE), includes receiving, from a network entity, a machine learning model for federated learning. The method also includes computing a set of gradient vector parameters during a first communication round of the federated learning for the machine learning model using a local dataset. The method further includes grouping the set of gradient vector parameters of the machine learning model into multiple subsets. The method also includes computing a representative value of all gradients within each of the subsets to obtain representative values for each of the subsets. The method includes transmitting the representative values to the network entity for the first communication round of the federated learning.

    PRECODING TECHNIQUES FOR WIRELESS COMMUNICATIONS
    22.
    发明公开
    PRECODING TECHNIQUES FOR WIRELESS COMMUNICATIONS 审中-公开

    公开(公告)号：US20230283332A1

    公开(公告)日：2023-09-07

    申请号：US18173482

    申请日：2023-02-23

    Applicant: QUALCOMM Incorporated

    Inventor： Saeid SAHRAEI ,  Gokul SRIDHARAN ,  Shimman Arvind PATEL ,  Krishna Kiran MUKKAVILLI ,  June NAMGOONG ,  Taesang YOO ,  Renqiu WANG

    IPC: H04B7/0456 ,  H04B7/06 ,  H04W72/23

    CPC classification number: H04B7/0465 ,  H04B7/0634 ,  H04W72/23

    Abstract: Methods, systems, and devices for wireless communications are described in which a base station may identify a null space matrix that lies within a null space of an effective channel matrix for communications between the base station and a user equipment (UE). An indication of the null space matrix may be provided to the UE, and the null space matrix used to determine modifications to a precoding matrix. The base station and UE may determine a redistribution matrix that provides a reduced variance of transmission powers for a number of transmission channels, where a product of the null space matrix and the redistribution matrix may be computed and added to the precoding matrix to generate a modified precoding matrix. The modified precoding matrix may be used to generate the communications from the base station and UE with reduced power variance across channels.

    DISTORTION PROBING REFERENCE SIGNAL CONFIGURATION
    26.
    发明公开
    DISTORTION PROBING REFERENCE SIGNAL CONFIGURATION 审中-公开

    公开(公告)号：US20230246775A1

    公开(公告)日：2023-08-03

    申请号：US18194479

    申请日：2023-03-31

    Applicant: QUALCOMM Incorporated

    Inventor： Hung Dinh LY ,  June Namgoong ,  Taesang YOO ,  Krishna Kiran MUKKAVILLI ,  Naga BHUSHAN ,  Tingfang Ji ,  Hwan Joon Kwon ,  Pavan Kumar VITTHALADEVUNI ,  Jay Kumar SUNDARARAJAN

    IPC: H04L5/00 ,  H04B17/309

    CPC classification number: H04L5/0048 ,  H04B17/309

    Abstract: Methods, systems, and devices for wireless communications are described. A configuration for a reference signal used to determine a non-linear behavior of transmission components at a transmitting device may be determined. The configuration for the reference signal may be determined based on signaling transmitted by the transmitting device, signaling transmitted by a device that receives the reference signal, or both. Additionally, or alternatively, the configuration for the reference signal may be determined based on a configuration of other signals transmitted by the transmitting device prior to or concurrently with the transmission of the reference signal. The determined configuration may be used to generate and transmit the reference signal or to determine a configuration of a received reference signal. In both cases, a non-linear response of transmission components at the transmitting device may be determined based on the reference signal.

    USER EQUIPMENT PARTICIPATION INDICATIONS ASSOCIATED WITH FEDERATED LEARNING
    29.
    发明申请
    USER EQUIPMENT PARTICIPATION INDICATIONS ASSOCIATED WITH FEDERATED LEARNING 有权

    公开(公告)号：US20230080218A1

    公开(公告)日：2023-03-16

    申请号：US17447668

    申请日：2021-09-14

    Applicant: QUALCOMM Incorporated

    Inventor： Hamed PEZESHKI ,  Taesang YOO ,  Tao LUO ,  Eren BALEVI

    IPC: H04W72/04 ,  G06N5/02

    Abstract: Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a user equipment (UE) may receive, from a base station, a federated learning configuration that configures a participation indication to be used by the UE to indicate a participation status of the UE associated with at least one federated learning round corresponding to a machine learning component. The UE may transmit the participation indication to the base station based at least in part on the federated learning configuration. Numerous other aspects are described.

    GENERALIZED NEURAL NETWORK ARCHITECTURES BASED ON FREQUENCY AND/OR TIME DIVISION
    30.
    发明申请
    GENERALIZED NEURAL NETWORK ARCHITECTURES BASED ON FREQUENCY AND/OR TIME DIVISION 有权

    公开(公告)号：US20230007530A1

    公开(公告)日：2023-01-05

    申请号：US17365510

    申请日：2021-07-01

    Applicant: QUALCOMM Incorporated

    Inventor： Pavan Kumar VITTHALADEVUNI ,  Taesang YOO ,  Naga BHUSHAN

    IPC: H04W28/06 ,  H04W24/10 ,  G06N20/20

    Abstract: Certain aspects of the present disclosure provide techniques for measurement encoding and decoding using neural networks to compress and decompress measurement data. One example method generally includes: generating, via each of a plurality of neural network encoders operating on measurement data, a compressed measurement based on a respective portion of the measurement data, wherein each of the neural network encoders is based on the same neural network model; generating at least one message indicative of the measurement data based on the compressed measurements; and transmitting the at least one message.

Patent Agency Ranking