PRIVACY-AWARE PRUNING IN MACHINE LEARNING

    公开(公告)号:US20220318412A1

    公开(公告)日:2022-10-06

    申请号:US17223946

    申请日:2021-04-06

    Abstract: Certain aspects of the present disclosure provide techniques for improved machine learning using private variational dropout. A set of parameters of a global machine learning model is updated based on a local data set, and the set of parameters is pruned based on pruning criteria. A noise-augmented set of gradients is computed for a subset of parameters remaining after the pruning, based in part on a noise value, and the noise-augmented set of gradients is transmitted to a global model server.

    SUPPORTING RANDOM ACCESS TYPE SELECTION BY A USER EQUIPMENT

    公开(公告)号:US20220210838A1

    公开(公告)日:2022-06-30

    申请号:US17606988

    申请日:2020-05-08

    Abstract: Methods, systems, and devices for wireless communications are described. Generally, the described techniques provide for a user equipment (UE) receiving a configuration message from a base station for supporting random access channel (RACH) type selection by the UE. The configuration message may include one or more reference signals and one or more link quality thresholds corresponding to the one or more reference signals. The UE may generate measurements of the reference signals and determine link quality for communications between the UE and the base station based on the measurements. Based on a comparison between the link quality to corresponding link quality thresholds, the UE may select a two-step random access procedure, a four-step random access procedure, or both for establishing a connection with the base station. In some cases, the UE considers system loading information, transmission parameters, or random access rules in selecting the random access procedure.

Patent Agency Ranking