METHOD OF GENERALIZED BALANCED FEW-SHOT LEARNING USING ONLY NOVEL DATA WITHOUT OLD DATA

    公开(公告)号:US20240135250A1

    公开(公告)日:2024-04-25

    申请号:US18356572

    申请日:2023-07-21

    IPC分类号: G06N20/00

    CPC分类号: G06N20/00

    摘要: A method and system of generalized balanced few-shot learning using only novel data without old data is proposed. The method of generalized balanced few-shot learning using only novel data without old data proposed in the present disclosure includes pre-training stage for training a feature extractor and a classifier of a training model with base data through a pre-training unit and fine-tuning stage for freezing the feature extractor through a fine-tuning unit, training a joint linear classifier capable of inferring base classes and novel classes, and performing weight normalization to achieve zero-mean and balanced variance.

    TERNARY NEURAL NETWORK ACCELERATOR DEVICE AND METHOD OF OPERATING THE SAME

    公开(公告)号:US20240037381A1

    公开(公告)日:2024-02-01

    申请号:US18303941

    申请日:2023-04-20

    IPC分类号: G06N3/065

    CPC分类号: G06N3/065

    摘要: According to one aspect of the present invention, a ternary neural network accelerator device includes a first semiconductor device comprising a first source terminal, a first drain terminal, and a first gate terminal, a second semiconductor device comprising a second source terminal, a second drain terminal, and a second gate terminal, a first searching line connected to the first drain terminal, a second searching line connected to the second drain terminal, and a matching line commonly connected to the first source terminal and the second source terminal, wherein ternary weight and ternary input are each set by either of a first operation and a second operation and nine computation results are output through the matching line according to conditions of the ternary weight and ternary input.

    IN-STORE AUTOMATIC PAYMENT METHOD, SYSTEM, AND PROGRAM

    公开(公告)号:US20230162267A1

    公开(公告)日:2023-05-25

    申请号:US17047868

    申请日:2020-05-08

    发明人: Phill Kyu RHEE

    摘要: The present disclosure relates to an in-store automatic payment method and system applied to an unmanned store service, wherein the method and the system reduce the burden of excessively collecting videos from a number of fixed cameras installed in the related art by using an egocentric video, computer computation overhead, and installation expense and operating expense of a system for an unmanned store. The method includes: collecting, by a mobile terminal, an egocentric video; detecting, by at least one among multiple devices, a purchase target item from the egocentric video; calculating, by at least one among the multiple devices, a level of reliability for the purchase target item from the egocentric video; and registering, by at least one among the multiple devices, the purchase target item on a selected-item list of a user when the level of reliability is equal to or greater than a preset threshold.