公开(公告)号：US12254064B2

公开(公告)日：2025-03-18

申请号：US17488735

申请日：2021-09-29

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Hanting Chen ,  Yunhe Wang ,  Chuanjian Liu ,  Kai Han ,  Chunjing Xu

IPC: G06F18/214 ,  G06F18/243 ,  G06N3/047

Abstract: The present application discloses an image generation method, a neural network compression method, and a related apparatus and device in the field of artificial intelligence. The image generation method includes: inputting a first matrix into an initial image generator to obtain a generated image; inputting the generated image into a preset discriminator to obtain a determining result, where the preset discriminator is obtained through training based on a real image and a category corresponding to the real image; updating the initial image generator based on the determining result to obtain a target image generator; and further inputting a second matrix into the target image generator to obtain a sample image. Further, a neural network compression method is disclosed, to compress the preset discriminator based on the sample image obtained by using the foregoing image generation method.

公开(公告)号：US20230206069A1

公开(公告)日：2023-06-29

申请号：US18175936

申请日：2023-02-28

Applicant: Huawei Technologies Co., Ltd.

Inventor： Junlei Zhang ,  Chuanjian Liu ,  Guilin Li ,  Xing Zhang ,  Wei Zhang ,  Zhenguo Li

IPC: G06N3/08 ,  G06N3/045

CPC classification number: G06N3/08 ,  G06N3/045

Abstract: A deep learning training method includes obtaining a training set, a first neural network, and a second neural network, where shortcut connections included in the first neural network are less than shortcut connections included in the second neural network; performing at least one time of iterative training on the first neural network based on the training set, to obtain a trained first neural network, where any iterative training includes: using a first output of at least one first intermediate layer in the first neural network as an input of at least one network layer in the second neural network, to obtain an output result of the at least one network layer; and updating the first neural network according to a first loss function.

公开(公告)号：US20250117637A1

公开(公告)日：2025-04-10

申请号：US18961921

申请日：2024-11-27

Applicant: Huawei Technologies Co., Ltd.

Inventor： Ying Nie ,  Kai Han ,  Chuanjian Liu ,  Junhui Ma ,  Yunhe Wang

IPC: G06N3/0495

Abstract: A neural network parameter quantization method includes obtaining a parameter of each neuron in a to-be-quantized model to obtain a parameter set, clustering parameters in the parameter set to obtain types of classified data, and quantizing each type of classified data in the types of classified data to obtain at least one type of quantization parameter, where the at least one type of quantization parameter is used to obtain a compression model, and precision of the at least one type of quantization parameter is lower than precision of a parameter in the to-be-quantized model.

公开(公告)号：US11450146B2

公开(公告)日：2022-09-20

申请号：US16776282

申请日：2020-01-29

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Liang Wang ,  Songcen Xu ,  Chuanjian Liu ,  Jun He

IPC: G06K9/00 ,  G06K9/62 ,  G06F3/01 ,  G06N3/08 ,  G06N20/20 ,  G06V40/20 ,  G06N20/10

Abstract: This application provides a gesture recognition method, and relates to the field of man-machine interaction technologies. The method includes: extracting M images from a first video segment in a video stream; performing gesture recognition on the M images by using a deep learning algorithm, to obtain a gesture recognition result corresponding to the first video segment; and performing result combination on gesture recognition results of N consecutive video segments including the first video segment, to obtain a combined gesture recognition result. In the foregoing recognition process, a gesture in the video stream does not need to be segmented or tracked, but phase actions are recognized by using a deep learning algorithm with a relatively fast calculation speed, and then the phase actions are combined, so as to improve a gesture recognition speed, and reduce a gesture recognition delay.

公开(公告)号：US20190156917A1

公开(公告)日：2019-05-23

申请号：US16251920

申请日：2019-01-18

Applicant: Huawei Technologies Co., Ltd.

Inventor： Chuanjian Liu ,  Liqun Deng ,  Guowei Huang

IPC: G16B30/10 ,  G16B40/10

Abstract: A data processing method includes traversing all sample fragments in a first sample set and collecting statistics about a first statistic of each basic element in a reference sample and included in the sample fragments, determining that a position of a basic element in the reference sample whose first statistic is less than a first threshold is a spacing position, dividing the reference sample into at least two reference sub-samples, traversing all the sample fragments in the first sample set and collecting statistics about a second statistic of each reference sub-sample of the reference sample and including the sample fragments, and combining adjacent reference sub-samples when a sum of second statistics of the adjacent reference sub-samples is less than a second threshold.