知搜-全球专利检索

  1. Login
  2. Sign Up

Search Results

180 records (took 0.027 seconds)

    MPE handling in over-the-air gradient vector aggregation for federated learning
    141.
    发明授权
    MPE handling in over-the-air gradient vector aggregation for federated learning 有权

    公开(公告)号：US11991644B2

    公开(公告)日：2024-05-21

    申请号：US17465625

    申请日：2021-09-02

    Applicant: QUALCOMM Incorporated

    Inventor： Hamed Pezeshki ,  Tao Luo ,  Sony Akkarakaran

    IPC: H04W52/22 ,  H04W52/14 ,  H04W52/36

    CPC classification number: H04W52/225 ,  H04W52/14 ,  H04W52/226 ,  H04W52/367

    Abstract: This disclosure provides systems, devices, apparatus, and methods, including computer programs encoded on storage media, for managing MPE conditions in over-the-air gradient vector aggregation for federated learning. A UE may receive a federated learning model from a base station for participating in a federated learning procedure with the base station. The UE may update the federated learning model for an iteration of the federated learning procedure based on whether an uplink transmission would cause an MPE threshold to be exceeded. The uplink transmission may have a Tx power level that corresponds to a configured Rx power level at the base station. The UE may drop the uplink transmission if the uplink transmission would cause the MPE threshold to be exceeded.

    LAYER-BY-LAYER TRAINING FOR FEDERATED LEARNING
    145.
    发明公开
    LAYER-BY-LAYER TRAINING FOR FEDERATED LEARNING 审中-公开

    公开(公告)号：US20230316062A1

    公开(公告)日：2023-10-05

    申请号：US17697751

    申请日：2022-03-17

    Applicant: QUALCOMM Incorporated

    Inventor： Eren Balevi ,  Taesang Yoo ,  Tao Luo ,  Srinivas Yerramalli ,  Junyl Li ,  Hamed Pezeshki

    IPC: H04B7/06 ,  G06N3/04 ,  G06N3/08

    CPC classification number: G06N3/08 ,  G06N3/04 ,  H04B7/0626

    Abstract: Methods, systems, and devices for wireless communications are described. A network entity may transmit an indication of neural network weights to one or more user equipments (UEs). The neural network weights may be for one or more shared layers of a federated learning neural network. The UEs may train a personalized layer of the neural network using the weights and data at the UEs. The UEs may transmit layer updates to the network entity. The network entity may train the neural network based on the updates. The UEs may send a transmission to the network entity that may be processed according to the neural network at the UEs and the network entity.

    REFERENCE SIGNAL PERIODICITIES FOR MULTIPLE SETS OF BEAMS
    150.
    发明公开
    REFERENCE SIGNAL PERIODICITIES FOR MULTIPLE SETS OF BEAMS 审中-公开

    公开(公告)号：US20230247458A1

    公开(公告)日：2023-08-03

    申请号：US17590653

    申请日：2022-02-01

    Applicant: QUALCOMM Incorporated

    Inventor： Hamed Pezeshki ,  Mahmoud Taherzadeh Boroujeni ,  Tao Luo

    IPC: H04W24/10 ,  H04L5/00 ,  H04W24/08 ,  H04B7/08

    CPC classification number: H04W24/10 ,  H04L5/0051 ,  H04W24/08 ,  H04B7/088

    Abstract: Methods, systems, and devices for wireless communications are described. A user equipment (UE) may receive control signaling from a network node indicating a reference signal (RS) configuration for RS measurements. The RS configuration may indicate a first set of beams, a first periodicity for measuring RSs on the first set of beams, a second set of beams, and a second periodicity for measuring RSs on the second set of beams. The UE may monitor a wireless channel for a first set of RSs communicated via the first set of beams according to the first periodicity and monitor the wireless channel for a second set of RSs communicated via the second set of beams according to the second periodicity. The UE and the network node may dynamically update one or more of the beam sets, periodicities, or both, to perform RS measurements.

Patent Agency Ranking