中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

48 records (took 0.019 seconds)

    SYSTEM AND METHOD FOR A UNIFIED ARCHITECTURE MULTI-TASK DEEP LEARNING MACHINE FOR OBJECT RECOGNITION
    13.
    发明申请
    SYSTEM AND METHOD FOR A UNIFIED ARCHITECTURE MULTI-TASK DEEP LEARNING MACHINE FOR OBJECT RECOGNITION 审中-公开

    公开(公告)号：US20180307897A1

    公开(公告)日：2018-10-25

    申请号：US16024823

    申请日：2018-06-30

    Applicant: Samsung Electronics Co., Ltd.

    Inventor： Mostafa EL-KHAMY ,  Arvind YEDLA ,  Marcel NASSAR ,  Jungwon LEE

    IPC: G06K9/00

    CPC classification number: G06K9/00268 ,  G06K9/00228 ,  G06T7/70 ,  G06T2207/20081 ,  G06T2207/20084 ,  G06T2207/30201

    Abstract: A system to recognize objects in an image includes an object detection network outputs a first hierarchical-calculated feature for a detected object. A face alignment regression network determines a regression loss for alignment parameters based on the first hierarchical-calculated feature. A detection box regression network determines a regression loss for detected boxes based on the first hierarchical-calculated feature. The object detection network further includes a weighted loss generator to generate a weighted loss for the first hierarchical-calculated feature, the regression loss for the alignment parameters and the regression loss of the detected boxes. A backpropagator backpropagates the generated weighted loss. A grouping network forms, based on the first hierarchical-calculated feature, the regression loss for the alignment parameters and the bounding box regression loss, at least one of a box grouping, an alignment parameter grouping, and a non-maximum suppression of the alignment parameters and the detected boxes.

    METHOD AND APPARATUS FOR NEURAL NETWORK QUANTIZATION
    14.
    发明申请
    METHOD AND APPARATUS FOR NEURAL NETWORK QUANTIZATION 审中-公开

    公开(公告)号：US20180107925A1

    公开(公告)日：2018-04-19

    申请号：US15433531

    申请日：2017-02-15

    Applicant: Samsung Electronics Co., Ltd.

    Inventor： Yoo Jin CHOI ,  Mostafa EL-KHAMY ,  Jungwon LEE

    IPC: G06N3/08 ,  G06F17/16

    CPC classification number: G06N3/08 ,  G06F17/16 ,  G06N3/063 ,  G06N3/082 ,  G06N3/084

    Abstract: Apparatuses and methods of manufacturing same, systems, and methods for performing network parameter quantization in deep neural networks are described. In one aspect, diagonals of a second-order partial derivative matrix (a Hessian matrix) of a loss function of network parameters of a neural network are determined and then used to weight (Hessian-weighting) the network parameters as part of quantizing the network parameters. In another aspect, the neural network is trained using first and second moment estimates of gradients of the network parameters and then the second moment estimates are used to weight the network parameters as part of quantizing the network parameters. In yet another aspect, network parameter quantization is performed by using an entropy-constrained scalar quantization (ECSQ) iterative algorithm. In yet another aspect, network parameter quantization is performed by quantizing the network parameters of all layers of a deep neural network together at once.

    MULTI-EXPERT ADVERSARIAL REGULARIZATION FOR ROBUST AND DATA-EFFICIENT DEEP SUPERVISED LEARNING
    17.
    发明申请
    MULTI-EXPERT ADVERSARIAL REGULARIZATION FOR ROBUST AND DATA-EFFICIENT DEEP SUPERVISED LEARNING 有权

    公开(公告)号：US20220301296A1

    公开(公告)日：2022-09-22

    申请号：US17674832

    申请日：2022-02-17

    Applicant: Samsung Electronics Co., Ltd.

    Inventor： Behnam BABAGHOLAMI MOHAMADABADI ,  Qingfeng LIU ,  Mostafa EL-KHAMY ,  Jungwon LEE

    IPC: G06V10/82 ,  G06N5/04 ,  G06V10/778 ,  G06V10/776 ,  G06V10/774

    Abstract: A system and a method to train a neural network are disclosed. A first image is weakly and strongly augmented. The first image, the weakly and strongly augmented first images are input into a feature extractor to obtain augmented features. Each weakly augmented first image is input to a corresponding first expert head to determine a supervised loss for each weakly augmented first image. Each strongly augmented first image is input to a corresponding second expert head to determine a diversity loss for each strongly augmented first image. The feature extractor is trained to minimize the supervised loss on weakly augmented first images and to minimize a multi-expert consensus loss on strongly augmented first images. Each first expert head is trained to minimize the supervised loss for each weakly augmented first image, and each second expert head is trained to minimize the diversity loss for each strongly augmented first image.

    SYSTEM AND METHOD FOR FEDERATED LEARNING USING WEIGHT ANONYMIZED FACTORIZATION
    19.
    发明申请
    SYSTEM AND METHOD FOR FEDERATED LEARNING USING WEIGHT ANONYMIZED FACTORIZATION 有权

    公开(公告)号：US20210374608A1

    公开(公告)日：2021-12-02

    申请号：US17148557

    申请日：2021-01-13

    Applicant: Samsung Electronics Co., Ltd.

    Inventor： Mostafa EL-KHAMY ,  Jungwon LEE ,  Weituo HAO ,  Lawrence CARIN ,  Nikhil MEHTA ,  Kevin J. LIANG

    IPC: G06N20/00 ,  G06F16/23

    Abstract: A federated machine-learning system includes a global server and client devices. The server receives updates of weight factor dictionaries and factor strengths vectors from the clients, and generates a globally updated weight factor dictionary and a globally updated factor strengths vector. A client device selects a group of parameters from a global group of parameters, and trains a model using a dataset of the client device and the group of selected parameters. The client device sends to the server a client-updated weight factor dictionary and a client-updated factor strengths vector. The client device receives the globally updated weight factor dictionary and the globally updated factor strengths vector, and retrains the model using the dataset of the client device, the group of parameters selected by the client device, and the globally updated weight factor dictionary and the globally updated factor strengths vector.

Patent Agency Ranking